Hi everyone, me again - and with more questions, too! Get a cup of
tea, this might take a while...
I thought it might be useful to share some of the research that we've
been doing in this area at the BBC. I know I was surprised by some of
the findings, and while many of you are probably very familiar with
this stuff, others might find it useful.
The first thing we looked at was curriculum theory. There are, it
turns out, several competing theories of curriculum design in the
world. By far the two most common are what I’ll call:
a) 'Learning Outcomes'
b) 'Syllabus'
In the English speaking (and parts of the developed) world, 'learning
outcomes' is currently much more common. In the rest of the world,
'syllabus' is.
So - is there a difference, and does it matter?
There is, and for the purposes of good metadata, it does.
'Learning outcomes'
‘Learning outcomes’ are statements of what students should be able to
do - and prove that they can do - after a learning experience.
Outcomes must be measurable and specific. The four UK national
curricula are outcomes-based, and the USA's 'learning standards' (e.g.
Common Core) are also learning outcomes. The LRMI term
'competency' (‘competency’ which is definable as ‘being able to do
something’) is too.
Learning outcome statements are verb-oriented (I’m not sure what the
grammatically correct term for that is!) Here are some examples of
learning outcomes:
e.g English National Curriculum, KS1, En1 Speaking (Sample)
1. To speak clearly, fluently and confidently to different people,
pupils should be taught to:
a. speak with clear diction and appropriate intonation
b. choose words with precision
c. organise what they say
d. focus on the main point(s)
e. include relevant detail
f. take into account the needs of their listeners.
e.g. Common Core State Standards, English Language Arts Standards »
Speaking & Listening » Grade 1 (sample)
SL.1.1. Participate in collaborative conversations with diverse
partners about grade 1 topics and texts with peers and adults in small
and larger groups.
SL1.2. Follow agreed-upon rules for discussions (e.g., listening to
others with care, speaking one at a time about the topics and texts
under discussion).
SL1.3. Build on others’ talk in conversations by responding to the
comments of others through multiple exchanges.
SL1.4. Ask questions to clear up any confusion about the topics and
texts under discussion.
‘Syllabus’
Syllabus is the model that many of us grew up with, and what most
people think of first when they think of a ‘curriculum’. A syllabus is
intrinsically content-oriented. The focus is on what students should
be taught, not what they should be able to do after they have been
taught (the ‘outcomes’). Fundamentally, a syllabus model is a list of
topics, sub-topics, and (where required) descriptions. It can be - but
does not have to be - hierarchical.
Unfortunately, from an information modeling perspective, the ‘learning
outcomes’ curriculum model and the ‘syllabus’ curriculum model are not
actually compatible. For example, a student cannot demonstrate that
they can ‘cell structure’ because ‘cell structure’ isn’t an outcome in
itself.
How do these curriculum models work with what we know about teacher
and learner search behaviour?
We know that, faced with a search box like google’s (and that's what
we're designing for!), teachers and learners (well, all users really)
will search by keywords.
1. Social studies
2. Health
3. Math
4. Lesson plans
5. Career
6. Technology
7. Science
8. Reading
9. Spanish
10. Music
11. English
12. Marketing
13. Physical Education
14. Fractions
15. Grammar
16. Gravitational field strength
17. Poetry
18. Weather
19. Writing
20. Algebra
I’ve seen very similar results from klascement.net (in Dutch), our own
results, and a couple of others too.
The thing worth noting here is that none of the dominant search
strings are learning outcomes. They might be terms that can be found
in outcomes, but they are not themselves outcomes. (A learner can’t
demonstrate their ability to ‘Poetry’ because ‘poetry’ is not an
outcome.)
It is also worth noting there are no verbs in these strings, where all
strings describing learning outcomes must by definition contain
verbs.
Learning outcomes must also, because they must be grammatically
complete sentences, be long. The longest search term in the top 20
here is three words, ‘gravitational field strength’, but all the
outcomes in the sample (which are representative of the NC) are 10+
words long.
So, even if the markup is accurately populated with complete outcomes,
matches between the (long) outcome and the (short) search term will be
a bit of a kludge.
I find it hard to imagine any teacher or learner ever searching on the
string ‘Build on others’ talk in conversations by responding to the
comments of others through multiple exchanges.’
This leads us to a strange and (to my mind) unexpected paradox - an
accurate description of a learning outcome does not equal a good
searchable string for those seeking to teach to the outcome.
This isn’t too surprising when assessed against the two curriculum
models - learning resources are made of content, and a syllabus model
is geared towards content. An outcomes model is not geared towards
content, it just uses content as a means to get to the outcome - so it
makes a bad match with searching.
Outcomes don't give good keywords, either.In our experience so far at
the BBC, learning outcomes as written in national curricula are either
very difficult or impossible to derive good keywords from (as the
person who had to try to do this with the four current UK national
curricula I’m quite confident in this assertion, but I’m happy to
discuss it if anyone wants clarification).
It is also worth noting that all the outcomes based curricula we’ve
seen are written in such a way that many, many resources could be used
to fulfil them - for the outcome ‘Build on others’ talk in
conversations by responding to the comments of others through multiple
exchanges’, the topic of conversation could be anything. Marking up a
given resource (be it about rabbits or the rainforest) as being
appropriate for this outcome would be accurate, but it would not
necessarily be useful.
There is good news, though!
The same source as provided the top twenty has recently introduced
subject (Maths, Science, etc.) filters, and they have been hugely
popular. So we know that educators really do value - and use -
semantic markup.
The question then becomes, if learning outcomes don’t create good/
easily seachable markup, then what does?
There are some clues in the top twenty that I think are worth
considering.
1. ‘Subject’ as a specific education-meaningful class (Maths, Science,
English etc). The evidence shows these to be very popular search terms
for educators. They are also ambiguous as non-semantic strings -
‘English’ the school subject is different from ‘English’ the language
(itself a viable subject in schema.org 'Thing' markup) and ‘English’
the adjective, ‘Science’ the school subject is different from ‘Science
+fiction’. We think that this area is ripe for semantic markup, but as
importantly, we’ve come to the conclusion that the real-world
education-specific infrastructure of ‘subject’ is worth maintaining in
our data architecture: ‘schoolSubject’ is different from ‘subject’
because schoolSubject is a complete, domain-specific administrative
group, making it a homograph of but not a synonym with ‘subject-as-
topic’. Perhaps introducing ‘Discipline’ as a LRMI field, external to
the Thing schema’s ‘subject’, would be useful in reflecting this
division.
2. All the learning outcomes have one extraordinarily useful
attribute: they exist in the context of either a level (e.g. a Key
Stage) and/or a qualification (GCSE, SATs, NVQ, IELTS, UAI/HSC/VSCE if
you’re an Australian like me and so on). The outcomes themselves are
at their most meaningful within the context of the level/
qualification. As a searchable string, ‘GCSE’ will have a much higher
level of useable meaning than ‘16’ - especially given that the LRMI
has been arranged to collect ‘learning outcomes’ in the ‘competency’
field.
3. I’m sure we’ve all been glad to see the Common Core standards gain
traction, and I am especially glad to see them pick up on the ASN’s
work by introducing Official Identifiers. It’s almost inevitable that
standards worldwide will start to move towards machine readability,
and I think the LRMI is in an excellent position to anticipate and
facilitate that shift. If conjunction with ‘Qualification’, an
‘identifier’ field would make a huge impact on this (and would be, I
think, far more robust than full-text retyping of each outcome).
4. We’re very, very lucky - our user population is highly skilled.
Most of them will have a very good idea of what they want their
students to be able to do at the end of a learning experience, so the
weight of marking up by outcome is reduced for us. (I also note that
outcomes have turnover - they only last about five years before
...
This discussion about the searchability of learning outcomes vs.
subject/discipline is really interesting. Is this on your blog somewhere,
where we can point others to it?
On Fri, Apr 27, 2012 at 11:36 AM, Zoe <zoe.f.r...@gmail.com> wrote:
> Hi everyone, me again - and with more questions, too! Get a cup of
> tea, this might take a while...
> I thought it might be useful to share some of the research that we've
> been doing in this area at the BBC. I know I was surprised by some of
> the findings, and while many of you are probably very familiar with
> this stuff, others might find it useful.
> The first thing we looked at was curriculum theory. There are, it
> turns out, several competing theories of curriculum design in the
> world. By far the two most common are what I’ll call:
> a) 'Learning Outcomes'
> b) 'Syllabus'
> In the English speaking (and parts of the developed) world, 'learning
> outcomes' is currently much more common. In the rest of the world,
> 'syllabus' is.
> So - is there a difference, and does it matter?
> There is, and for the purposes of good metadata, it does.
> 'Learning outcomes'
> ‘Learning outcomes’ are statements of what students should be able to
> do - and prove that they can do - after a learning experience.
> Outcomes must be measurable and specific. The four UK national
> curricula are outcomes-based, and the USA's 'learning standards' (e.g.
> Common Core) are also learning outcomes. The LRMI term
> 'competency' (‘competency’ which is definable as ‘being able to do
> something’) is too.
> Learning outcome statements are verb-oriented (I’m not sure what the
> grammatically correct term for that is!) Here are some examples of
> learning outcomes:
> e.g English National Curriculum, KS1, En1 Speaking (Sample)
> 1. To speak clearly, fluently and confidently to different people,
> pupils should be taught to:
> a. speak with clear diction and appropriate intonation
> b. choose words with precision
> c. organise what they say
> d. focus on the main point(s)
> e. include relevant detail
> f. take into account the needs of their listeners.
> e.g. Common Core State Standards, English Language Arts Standards »
> Speaking & Listening » Grade 1 (sample)
> SL.1.1. Participate in collaborative conversations with diverse
> partners about grade 1 topics and texts with peers and adults in small
> and larger groups.
> SL1.2. Follow agreed-upon rules for discussions (e.g., listening to
> others with care, speaking one at a time about the topics and texts
> under discussion).
> SL1.3. Build on others’ talk in conversations by responding to the
> comments of others through multiple exchanges.
> SL1.4. Ask questions to clear up any confusion about the topics and
> texts under discussion.
> ‘Syllabus’
> Syllabus is the model that many of us grew up with, and what most
> people think of first when they think of a ‘curriculum’. A syllabus is
> intrinsically content-oriented. The focus is on what students should
> be taught, not what they should be able to do after they have been
> taught (the ‘outcomes’). Fundamentally, a syllabus model is a list of
> topics, sub-topics, and (where required) descriptions. It can be - but
> does not have to be - hierarchical.
> Unfortunately, from an information modeling perspective, the ‘learning
> outcomes’ curriculum model and the ‘syllabus’ curriculum model are not
> actually compatible. For example, a student cannot demonstrate that
> they can ‘cell structure’ because ‘cell structure’ isn’t an outcome in
> itself.
> How do these curriculum models work with what we know about teacher
> and learner search behaviour?
> We know that, faced with a search box like google’s (and that's what
> we're designing for!), teachers and learners (well, all users really)
> will search by keywords.
> 1. Social studies
> 2. Health
> 3. Math
> 4. Lesson plans
> 5. Career
> 6. Technology
> 7. Science
> 8. Reading
> 9. Spanish
> 10. Music
> 11. English
> 12. Marketing
> 13. Physical Education
> 14. Fractions
> 15. Grammar
> 16. Gravitational field strength
> 17. Poetry
> 18. Weather
> 19. Writing
> 20. Algebra
> I’ve seen very similar results from klascement.net (in Dutch), our own
> results, and a couple of others too.
> The thing worth noting here is that none of the dominant search
> strings are learning outcomes. They might be terms that can be found
> in outcomes, but they are not themselves outcomes. (A learner can’t
> demonstrate their ability to ‘Poetry’ because ‘poetry’ is not an
> outcome.)
> It is also worth noting there are no verbs in these strings, where all
> strings describing learning outcomes must by definition contain
> verbs.
> Learning outcomes must also, because they must be grammatically
> complete sentences, be long. The longest search term in the top 20
> here is three words, ‘gravitational field strength’, but all the
> outcomes in the sample (which are representative of the NC) are 10+
> words long.
> So, even if the markup is accurately populated with complete outcomes,
> matches between the (long) outcome and the (short) search term will be
> a bit of a kludge.
> I find it hard to imagine any teacher or learner ever searching on the
> string ‘Build on others’ talk in conversations by responding to the
> comments of others through multiple exchanges.’
> This leads us to a strange and (to my mind) unexpected paradox - an
> accurate description of a learning outcome does not equal a good
> searchable string for those seeking to teach to the outcome.
> This isn’t too surprising when assessed against the two curriculum
> models - learning resources are made of content, and a syllabus model
> is geared towards content. An outcomes model is not geared towards
> content, it just uses content as a means to get to the outcome - so it
> makes a bad match with searching.
> Outcomes don't give good keywords, either.In our experience so far at
> the BBC, learning outcomes as written in national curricula are either
> very difficult or impossible to derive good keywords from (as the
> person who had to try to do this with the four current UK national
> curricula I’m quite confident in this assertion, but I’m happy to
> discuss it if anyone wants clarification).
> It is also worth noting that all the outcomes based curricula we’ve
> seen are written in such a way that many, many resources could be used
> to fulfil them - for the outcome ‘Build on others’ talk in
> conversations by responding to the comments of others through multiple
> exchanges’, the topic of conversation could be anything. Marking up a
> given resource (be it about rabbits or the rainforest) as being
> appropriate for this outcome would be accurate, but it would not
> necessarily be useful.
> There is good news, though!
> The same source as provided the top twenty has recently introduced
> subject (Maths, Science, etc.) filters, and they have been hugely
> popular. So we know that educators really do value - and use -
> semantic markup.
> The question then becomes, if learning outcomes don’t create good/
> easily seachable markup, then what does?
> There are some clues in the top twenty that I think are worth
> considering.
> 1. ‘Subject’ as a specific education-meaningful class (Maths, Science,
> English etc). The evidence shows these to be very popular search terms
> for educators. They are also ambiguous as non-semantic strings -
> ‘English’ the school subject is different from ‘English’ the language
> (itself a viable subject in schema.org 'Thing' markup) and ‘English’
> the adjective, ‘Science’ the school subject is different from ‘Science
> +fiction’. We think that this area is ripe for semantic markup, but as
> importantly, we’ve come to the conclusion that the real-world
> education-specific infrastructure of ‘subject’ is worth maintaining in
> our data architecture: ‘schoolSubject’ is different from ‘subject’
> because schoolSubject is a complete, domain-specific administrative
> group, making it a homograph of but not a synonym with ‘subject-as-
> topic’. Perhaps introducing ‘Discipline’ as a LRMI field, external to
> the Thing schema’s ‘subject’, would be useful in reflecting this
> division.
> 2. All the learning outcomes have one extraordinarily useful
> attribute: they exist in the context of either a level (e.g. a Key
> Stage) and/or a qualification (GCSE, SATs, NVQ, IELTS, UAI/HSC/VSCE if
> you’re an Australian like me and so on). The outcomes themselves are
> at their most meaningful within the context of the level/
> qualification. As a searchable string, ‘GCSE’ will have a much higher
> level of useable meaning than ‘16’ - especially given that the LRMI
> has been arranged to collect ‘learning outcomes’ in the ‘competency’
> field.
> 3. I’m sure we’ve all been glad to see the Common Core standards gain
> traction, and I am especially glad to see them pick up on the ASN’s
> work by introducing Official Identifiers. It’s almost inevitable that
> standards worldwide will start
> This discussion about the searchability of learning outcomes vs. > subject/discipline is really interesting. Is this on your blog > somewhere, where we can point others to it?
-- Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012
We invite research leaders and ambitious early career
researchers to join us in leading and driving research
in key inter-disciplinary themes. Please see
You cover a lot of ground here, and there is much we might dig into.
However, for the pending work to finalize 1.0 of LRMI, let me point to a
couple of important implications.
First you again demonstrate that "Competencies" may not be an appropriately
flexible term to represent the relationship between a resource and the two
competing structural frameworks for organizing curriculum design you discuss
(Outcomes vs. Syllabus). There are others we might list as well. All of them
seem to boil down to some form of catalog, be it of skills or topics to be
covered. I will note here that in the US, the K-12 learning standards are
met via programs of instruction that are guided by an "instructional scope
and sequence" which more closely resembles a Syllabus. Alignment to a Scope
and Sequence document is another kind of curriculum alignment.
If we put the term Competency aside, and keep the same structure, we really
are defining an alignment or an association between some resource and some
framework (Be it a skill, or content framework). This alignment has some
number of classifications (AlignmentTypes), but what those are (Teaches,
Assesses, Requires, etc.) really depends on the thing being aligned, and the
framework it is being aligned to. The same resource ("how to get dressed"
for example), might require a skill ("can tie shoes") and address part of a
syllabus topic ("life skills > things you do in the morning"). I believe
the basic structure provided enables both types of alignment, really.
Another is the ReadingLevel case, which aligns a text document to a reading
complexity scale or framework.
So the second thing you point to, indirectly is the need to support
different classifications of alignment.
The last important point I will call out is your observation that similar
things can and are often described in very different ways depending on
content and use. This I fear is not really a solvable issue. But as you
observe, if everyone simply tagged their learning resources as being about
some topic at some level for some type of user an aligned to some structured
topical or skill based framework, we will be much better of then we are
today. If we happen to use the same terms and frameworks, that would be even
better, but I won't be holding my breath on that one.
Joshua Marks
CTO
Curriki: The Global Education and Learning Community jma...@curriki.org
www.curriki.org
I welcome you to become a member of the Curriki community, to follow us
on Twitter and to say hello on our blog, Facebook and LinkedIn communities.
-----Original Message-----
From: lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] On Behalf Of Zoe
Sent: Friday, April 27, 2012 9:36 AM
To: Learning Resource Metadata Initiative
Subject: Some thoughts from our research on curriculum models in metadata
Hi everyone, me again - and with more questions, too! Get a cup of tea, this
might take a while...
I thought it might be useful to share some of the research that we've been
doing in this area at the BBC. I know I was surprised by some of the
findings, and while many of you are probably very familiar with this stuff,
others might find it useful.
The first thing we looked at was curriculum theory. There are, it turns out,
several competing theories of curriculum design in the world. By far the two
most common are what I’ll call:
a) 'Learning Outcomes'
b) 'Syllabus'
In the English speaking (and parts of the developed) world, 'learning
outcomes' is currently much more common. In the rest of the world,
'syllabus' is.
So - is there a difference, and does it matter?
There is, and for the purposes of good metadata, it does.
'Learning outcomes'
‘Learning outcomes’ are statements of what students should be able to do -
and prove that they can do - after a learning experience.
Outcomes must be measurable and specific. The four UK national curricula are
outcomes-based, and the USA's 'learning standards' (e.g.
Common Core) are also learning outcomes. The LRMI term 'competency'
(‘competency’ which is definable as ‘being able to do
something’) is too.
Learning outcome statements are verb-oriented (I’m not sure what the
grammatically correct term for that is!) Here are some examples of learning
outcomes:
e.g English National Curriculum, KS1, En1 Speaking (Sample)
1. To speak clearly, fluently and confidently to different people, pupils
should be taught to:
a. speak with clear diction and appropriate intonation b. choose words with
precision c. organise what they say d. focus on the main point(s) e. include
relevant detail f. take into account the needs of their listeners.
e.g. Common Core State Standards, English Language Arts Standards » Speaking
& Listening » Grade 1 (sample)
SL.1.1. Participate in collaborative conversations with diverse partners
about grade 1 topics and texts with peers and adults in small and larger
groups.
SL1.2. Follow agreed-upon rules for discussions (e.g., listening to others
with care, speaking one at a time about the topics and texts under
discussion).
SL1.3. Build on others’ talk in conversations by responding to the comments
of others through multiple exchanges.
SL1.4. Ask questions to clear up any confusion about the topics and texts
under discussion.
‘Syllabus’
Syllabus is the model that many of us grew up with, and what most people
think of first when they think of a ‘curriculum’. A syllabus is
intrinsically content-oriented. The focus is on what students should be
taught, not what they should be able to do after they have been taught (the
‘outcomes’). Fundamentally, a syllabus model is a list of topics,
sub-topics, and (where required) descriptions. It can be - but does not have
to be - hierarchical.
Unfortunately, from an information modeling perspective, the ‘learning
outcomes’ curriculum model and the ‘syllabus’ curriculum model are not
actually compatible. For example, a student cannot demonstrate that they can
‘cell structure’ because ‘cell structure’ isn’t an outcome in itself.
How do these curriculum models work with what we know about teacher and
learner search behaviour?
We know that, faced with a search box like google’s (and that's what we're
designing for!), teachers and learners (well, all users really) will search
by keywords.
1. Social studies
2. Health
3. Math
4. Lesson plans
5. Career
6. Technology
7. Science
8. Reading
9. Spanish
10. Music
11. English
12. Marketing
13. Physical Education
14. Fractions
15. Grammar
16. Gravitational field strength
17. Poetry
18. Weather
19. Writing
20. Algebra
I’ve seen very similar results from klascement.net (in Dutch), our own
results, and a couple of others too.
The thing worth noting here is that none of the dominant search strings are
learning outcomes. They might be terms that can be found in outcomes, but
they are not themselves outcomes. (A learner can’t demonstrate their ability
to ‘Poetry’ because ‘poetry’ is not an
outcome.)
It is also worth noting there are no verbs in these strings, where all
strings describing learning outcomes must by definition contain verbs.
Learning outcomes must also, because they must be grammatically complete
sentences, be long. The longest search term in the top 20 here is three
words, ‘gravitational field strength’, but all the outcomes in the sample
(which are representative of the NC) are 10+ words long.
So, even if the markup is accurately populated with complete outcomes,
matches between the (long) outcome and the (short) search term will be a bit
of a kludge.
I find it hard to imagine any teacher or learner ever searching on the
string ‘Build on others’ talk in conversations by responding to the comments
of others through multiple exchanges.’
This leads us to a strange and (to my mind) unexpected paradox - an accurate
description of a learning outcome does not equal a good searchable string
for those seeking to teach to the outcome.
This isn’t too surprising when assessed against the two curriculum models -
learning resources are made of content, and a syllabus model is geared
towards content. An outcomes model is not geared towards content, it just
uses content as a means to get to the outcome - so it makes a bad match with
searching.
Outcomes don't give good keywords, either.In our experience so far at the
BBC, learning outcomes as written in national curricula are either very
difficult or impossible to derive good keywords from (as the person who had
to try to do this with the four current UK national curricula I’m quite
confident in this assertion, but I’m happy to discuss it if anyone wants
clarification).
It is also worth noting that all the outcomes based curricula we’ve seen are
written in such a way that many, many resources could be used to fulfil them
- for the outcome ‘Build on others’ talk in conversations by responding to
the comments of others through multiple exchanges’, the topic of
conversation could be anything. Marking up a given resource (be it about
rabbits or the rainforest) as being appropriate for this outcome would be
accurate, but it would not necessarily be useful.
Glad to see people are interested in these issues :)
Two posts coming: This one is about a theory point, the next will
respond directly to Josh.
Teasing out some big concepts from Josh's post - I could be wrong, but
I think I've spotted a theoretical point worth fleshing out. It has to
do with the usefulness of accuracy in a keyword-search context.
Competency, timeRequired, and interactivityType are the fields
affected.
The key take-home is: Accurate descriptions that work in catalogues
don't necessarily add value in keyword search. I think
(As an early-career cataloguing fiend, accuracy was my first love -
but I have, grudgingly, come to respect the other flavours of icecream
available...)
What I like about schema.org as opposed to, for example, our old &
trusted friend Dublin Core is it's usefulness to the search box. No
filters on the UI, no taxonomy, no controlled vocabs, just keyword
search.
That does mean, however, that anything stored in the LRMI has to be
something someone would use in a keyword search.
(Keyword search does not/never will provide as much accuracy/control
as a good catalogue record (most websites don't even have dates on
them, after all!) but it does what it does very well, if more
messily.)
Here are some terms that I can't imagine an educational-resource-
seeking user typing into google, either with or without other strings:
Assesses
Teaches
Active
Expositive
Mixed
P1H30M
For example: I can't imagine a teacher looking for Jane Eyre resources
typing 'Jane Eyre expositive'.
Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
'Jane Eyre essay questions' I can imagine though (I've searched that
one myself! I was an English teacher in a former life :) ), which is
nicely facilitated by LRMI's learningResourceType. I can also imagine
'Jane Eyre lesson plan', 'Jane Eyre GCSE', and (at a stretch) 'Jane
Eyre classroom activity one hour'.
What's interesting is that 'active', 'expositive', 'mixed', 'teaches',
'assesses', and ISO 8601 time forms are all very definitely *accurate*
(or more specifically, capable of providing accurate information). In
a classical cataloguing environement, they're very useful.
Within a google context, however, the story may be different. These
strings don't match with what we know about how people use that big
empty search box - the search box invites concrete, discrete concepts,
not context-dependent abstractions like 'mixed'.
This is another reason, I think, to make sure concrete concepts (a
couple I threw into the pot were were 'discipline' and
'qualification') have the chance to be included in the markup. What do
others think?
Zoe
On Apr 27, 11:20 pm, "Joshua Marks" <jma...@curriki.org> wrote:
> You cover a lot of ground here, and there is much we might dig into.
> However, for the pending work to finalize 1.0 of LRMI, let me point to a
> couple of important implications.
> First you again demonstrate that "Competencies" may not be an appropriately
> flexible term to represent the relationship between a resource and the two
> competing structural frameworks for organizing curriculum design you discuss
> (Outcomes vs. Syllabus). There are others we might list as well. All of them
> seem to boil down to some form of catalog, be it of skills or topics to be
> covered. I will note here that in the US, the K-12 learning standards are
> met via programs of instruction that are guided by an "instructional scope
> and sequence" which more closely resembles a Syllabus. Alignment to a Scope
> and Sequence document is another kind of curriculum alignment.
> If we put the term Competency aside, and keep the same structure, we really
> are defining an alignment or an association between some resource and some
> framework (Be it a skill, or content framework). This alignment has some
> number of classifications (AlignmentTypes), but what those are (Teaches,
> Assesses, Requires, etc.) really depends on the thing being aligned, and the
> framework it is being aligned to. The same resource ("how to get dressed"
> for example), might require a skill ("can tie shoes") and address part of a
> syllabus topic ("life skills > things you do in the morning"). I believe
> the basic structure provided enables both types of alignment, really.
> Another is the ReadingLevel case, which aligns a text document to a reading
> complexity scale or framework.
> So the second thing you point to, indirectly is the need to support
> different classifications of alignment.
> The last important point I will call out is your observation that similar
> things can and are often described in very different ways depending on
> content and use. This I fear is not really a solvable issue. But as you
> observe, if everyone simply tagged their learning resources as being about
> some topic at some level for some type of user an aligned to some structured
> topical or skill based framework, we will be much better of then we are
> today. If we happen to use the same terms and frameworks, that would be even
> better, but I won't be holding my breath on that one.
> Joshua Marks
> CTO
> Curriki: The Global Education and Learning Community
> jma...@curriki.orgwww.curriki.org
> I welcome you to become a member of the Curriki community, to follow us
> on Twitter and to say hello on our blog, Facebook and LinkedIn communities.
> -----Original Message-----
> From: lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] On Behalf Of Zoe
> Sent: Friday, April 27, 2012 9:36 AM
> To: Learning Resource Metadata Initiative
> Subject: Some thoughts from our research on curriculum models in metadata
> Hi everyone, me again - and with more questions, too! Get a cup of tea, this
> might take a while...
> I thought it might be useful to share some of the research that we've been
> doing in this area at the BBC. I know I was surprised by some of the
> findings, and while many of you are probably very familiar with this stuff,
> others might find it useful.
> The first thing we looked at was curriculum theory. There are, it turns out,
> several competing theories of curriculum design in the world. By far the two
> most common are what I’ll call:
> a) 'Learning Outcomes'
> b) 'Syllabus'
> In the English speaking (and parts of the developed) world, 'learning
> outcomes' is currently much more common. In the rest of the world,
> 'syllabus' is.
> So - is there a difference, and does it matter?
> There is, and for the purposes of good metadata, it does.
> 'Learning outcomes'
> ‘Learning outcomes’ are statements of what students should be able to do -
> and prove that they can do - after a learning experience.
> Outcomes must be measurable and specific. The four UK national curricula are
> outcomes-based, and the USA's 'learning standards' (e.g.
> Common Core) are also learning outcomes. The LRMI term 'competency'
> (‘competency’ which is definable as ‘being able to do
> something’) is too.
> Learning outcome statements are verb-oriented (I’m not sure what the
> grammatically correct term for that is!) Here are some examples of learning
> outcomes:
> e.g English National Curriculum, KS1, En1 Speaking (Sample)
> 1. To speak clearly, fluently and confidently to different people, pupils
> should be taught to:
> a. speak with clear diction and appropriate intonation b. choose words with
> precision c. organise what they say d. focus on the main point(s) e. include
> relevant detail f. take into account the needs of their listeners.
> e.g. Common Core State Standards, English Language Arts Standards » Speaking
> & Listening » Grade 1 (sample)
> SL.1.1. Participate in collaborative conversations with diverse partners
> about grade 1 topics and texts with peers and adults in small and larger
> groups.
> SL1.2. Follow agreed-upon rules for discussions (e.g., listening to others
> with care, speaking one at a time about the topics and texts under
> discussion).
> SL1.3. Build on others’ talk in conversations by responding to the comments
> of others through multiple exchanges.
> SL1.4. Ask questions to clear up any confusion about the topics and texts
> under discussion.
> ‘Syllabus’
> Syllabus is the model that many of us grew up with, and what most people
> think of first when they think of a ‘curriculum’. A syllabus is
> intrinsically content-oriented. The focus is on what students should be
> taught, not what they should be able to do after they have been taught (the
> ‘outcomes’). Fundamentally, a syllabus model is a list of topics,
> sub-topics, and (where required) descriptions. It can be - but does not have
> to be - hierarchical.
> Unfortunately, from an information modeling perspective, the ‘learning
> outcomes’ curriculum model and the ‘syllabus’ curriculum model are not
> actually compatible. For example, a student cannot demonstrate that they can
> ‘cell structure’ because ‘cell structure’ isn’t an outcome in itself.
> How do these curriculum models work with what we know about teacher and
> learner search behaviour?
> We know that, faced with a search box like google’s (and that's what we're
> designing for!), teachers and learners (well, all users really) will search
> by keywords.
> As evidence for this, here are the top 20 searches from thegateway.org
> (traffic = about 500 000 a month, I
Josh, thanks for giving me an insight into how this work has built up
and what you expect from it.
1. I hadn't seen 'scope and sequence' - that looks much more concrete.
In England we are waiting on a new curriculum, to be delivered next
month, which we have been told will be more content-oriented... though
we won't know until we see it!
2. I see what you mean - yes, they cater to both cases. This looks to
me, however, a bit like CBT/SCORM type thinking, where the computer
has a role in deciding what the learner should do next. On the open
web (as per my last long post!) I'm not sure that function is as
useful as providing markup to teachers, and working on the assumption
that they can/will decide what the learner should do next.
As an example, I note that your example 'things you do in the morning'
does not actually contain the string 'tying shoes'. If I was teaching
this class, I'd already know that I was doing a unit on things you do
in the morning, and if I decided that a session on tying shoes was
right for my class, that's what I'd type into google - 'tying shoes'.
I guess what I'm saying is, google is not sequential. Google's just
one search, then another search, then another search. If we design for
sequencing, that is very useful for CBT programming in a closed
environment, but not for searchable html markup.
3. Ha ha - yes, I see. Our users search on simple strings, so (if I've
got you right) I agree - that's what we should be building for :)
Zoe
On Apr 27, 11:20 pm, "Joshua Marks" <jma...@curriki.org> wrote:
> You cover a lot of ground here, and there is much we might dig into.
> However, for the pending work to finalize 1.0 of LRMI, let me point to a
> couple of important implications.
> First you again demonstrate that "Competencies" may not be an appropriately
> flexible term to represent the relationship between a resource and the two
> competing structural frameworks for organizing curriculum design you discuss
> (Outcomes vs. Syllabus). There are others we might list as well. All of them
> seem to boil down to some form of catalog, be it of skills or topics to be
> covered. I will note here that in the US, the K-12 learning standards are
> met via programs of instruction that are guided by an "instructional scope
> and sequence" which more closely resembles a Syllabus. Alignment to a Scope
> and Sequence document is another kind of curriculum alignment.
> If we put the term Competency aside, and keep the same structure, we really
> are defining an alignment or an association between some resource and some
> framework (Be it a skill, or content framework). This alignment has some
> number of classifications (AlignmentTypes), but what those are (Teaches,
> Assesses, Requires, etc.) really depends on the thing being aligned, and the
> framework it is being aligned to. The same resource ("how to get dressed"
> for example), might require a skill ("can tie shoes") and address part of a
> syllabus topic ("life skills > things you do in the morning"). I believe
> the basic structure provided enables both types of alignment, really.
> Another is the ReadingLevel case, which aligns a text document to a reading
> complexity scale or framework.
> So the second thing you point to, indirectly is the need to support
> different classifications of alignment.
> The last important point I will call out is your observation that similar
> things can and are often described in very different ways depending on
> content and use. This I fear is not really a solvable issue. But as you
> observe, if everyone simply tagged their learning resources as being about
> some topic at some level for some type of user an aligned to some structured
> topical or skill based framework, we will be much better of then we are
> today. If we happen to use the same terms and frameworks, that would be even
> better, but I won't be holding my breath on that one.
> Joshua Marks
> CTO
> Curriki: The Global Education and Learning Community
> jma...@curriki.orgwww.curriki.org
> I welcome you to become a member of the Curriki community, to follow us
> on Twitter and to say hello on our blog, Facebook and LinkedIn communities.
> -----Original Message-----
> From: lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] On Behalf Of Zoe
> Sent: Friday, April 27, 2012 9:36 AM
> To: Learning Resource Metadata Initiative
> Subject: Some thoughts from our research on curriculum models in metadata
> Hi everyone, me again - and with more questions, too! Get a cup of tea, this
> might take a while...
> I thought it might be useful to share some of the research that we've been
> doing in this area at the BBC. I know I was surprised by some of the
> findings, and while many of you are probably very familiar with this stuff,
> others might find it useful.
> The first thing we looked at was curriculum theory. There are, it turns out,
> several competing theories of curriculum design in the world. By far the two
> most common are what I’ll call:
> a) 'Learning Outcomes'
> b) 'Syllabus'
> In the English speaking (and parts of the developed) world, 'learning
> outcomes' is currently much more common. In the rest of the world,
> 'syllabus' is.
> So - is there a difference, and does it matter?
> There is, and for the purposes of good metadata, it does.
> 'Learning outcomes'
> ‘Learning outcomes’ are statements of what students should be able to do -
> and prove that they can do - after a learning experience.
> Outcomes must be measurable and specific. The four UK national curricula are
> outcomes-based, and the USA's 'learning standards' (e.g.
> Common Core) are also learning outcomes. The LRMI term 'competency'
> (‘competency’ which is definable as ‘being able to do
> something’) is too.
> Learning outcome statements are verb-oriented (I’m not sure what the
> grammatically correct term for that is!) Here are some examples of learning
> outcomes:
> e.g English National Curriculum, KS1, En1 Speaking (Sample)
> 1. To speak clearly, fluently and confidently to different people, pupils
> should be taught to:
> a. speak with clear diction and appropriate intonation b. choose words with
> precision c. organise what they say d. focus on the main point(s) e. include
> relevant detail f. take into account the needs of their listeners.
> e.g. Common Core State Standards, English Language Arts Standards » Speaking
> & Listening » Grade 1 (sample)
> SL.1.1. Participate in collaborative conversations with diverse partners
> about grade 1 topics and texts with peers and adults in small and larger
> groups.
> SL1.2. Follow agreed-upon rules for discussions (e.g., listening to others
> with care, speaking one at a time about the topics and texts under
> discussion).
> SL1.3. Build on others’ talk in conversations by responding to the comments
> of others through multiple exchanges.
> SL1.4. Ask questions to clear up any confusion about the topics and texts
> under discussion.
> ‘Syllabus’
> Syllabus is the model that many of us grew up with, and what most people
> think of first when they think of a ‘curriculum’. A syllabus is
> intrinsically content-oriented. The focus is on what students should be
> taught, not what they should be able to do after they have been taught (the
> ‘outcomes’). Fundamentally, a syllabus model is a list of topics,
> sub-topics, and (where required) descriptions. It can be - but does not have
> to be - hierarchical.
> Unfortunately, from an information modeling perspective, the ‘learning
> outcomes’ curriculum model and the ‘syllabus’ curriculum model are not
> actually compatible. For example, a student cannot demonstrate that they can
> ‘cell structure’ because ‘cell structure’ isn’t an outcome in itself.
> How do these curriculum models work with what we know about teacher and
> learner search behaviour?
> We know that, faced with a search box like google’s (and that's what we're
> designing for!), teachers and learners (well, all users really) will search
> by keywords.
> 1. Social studies
> 2. Health
> 3. Math
> 4. Lesson plans
> 5. Career
> 6. Technology
> 7. Science
> 8. Reading
> 9. Spanish
> 10. Music
> 11. English
> 12. Marketing
> 13. Physical Education
> 14. Fractions
> 15. Grammar
> 16. Gravitational field strength
> 17. Poetry
> 18. Weather
> 19. Writing
> 20. Algebra
> I’ve seen very similar results from klascement.net (in Dutch), our own
> results, and a couple of others too.
> The thing worth noting here is that none of the dominant search strings are
> learning outcomes. They might be terms that can be found in outcomes, but
> they are not themselves outcomes. (A learner can’t demonstrate their ability
> to ‘Poetry’ because ‘poetry’ is not an
> outcome.)
> It is also worth noting there are no verbs in these strings, where all
> strings describing learning outcomes must by definition contain verbs.
> Learning outcomes must also, because they must be grammatically complete
> sentences, be long. The longest search term in the top 20 here is three
> For example: I can't imagine a teacher looking for Jane Eyre resources
> typing 'Jane Eyre expositive'.
> Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
Zoe, it appears that there there is an underlying assumption to this thread
that metadata utility is all about one aspect of "search": what terms get
typed into the browser. But there are other critical user behaviors with
search beyond determining the search string. An equally important matter
is whether the descriptions returned by whatever is typed into the search
box contain useful information of a kind sufficient to support making a
decision whether to retrieve. "P1H30M" or "Expositive" may not be terms a
teacher initially types into Google, but that bit of information might well
support a decision to retrieve or not when what is needed is a lesson plan
that can be usefully deployed in under an hour and is purely expositive.
This is particularly true when result sets are appropriately faceted.
So, "Jane Eyre lesson plan" may be what is typed as the first step in
search, but the other structured metadata likely determines whether what is
returned by the engine is retrieved. So, Zoe, I think you are likely
spot-on in your analysis if the sole purpose of structured LRMI and
schema.org metadata is the magic bullet of that evasive perfect string to
type into the search box. But, deciding which resources in the retrieved
list to actually retrieve is a function beyond the search string.
Stuart
-- Stuart A. Sutton,
CEO and Managing Director, Dublin Core Metadata Initiative
Associate Professor Emeritus, The Information School
University of Washington
Following up on Stuart's response, the model I've had in my head for
justifying LRMI (and schema.org in general) is the faceted search that
Google now supports around recipes.
I'm never sure if this works for everyone, but when I do a search now for
"potato salad recipe" in Google, I now get a set of facets on the left
that let me narrow down my search, for ingredients, cook time and calorie
count.
As a casual recipe searcher, I think I'd be unlikely to use a
search term "potato salad recipe with less than 300 calories" much less
"potato salad recipe with 200 calories." But having that metadata exposed
to Google allows it to provide a better search experience.
You can imagine how this might work for LRMI metadata. The user starts
with a search on "Math," but can quickly see what the range of grade
levels, type of content, duration, etc are. She can then use a filter to
narrow the results, or perhaps realize she needs a more specific search
term.
I still believe full text searches are where most users want to start, but
I think there's a lot of room to enhance the search experience with better
metadata.
- Peter
Peter Pinch | Production Manager, OpenCourseWare
Massachusetts Institute of Technology
From: Stuart Sutton <stuartasut...@gmail.com>
Reply-To: <lrmi@googlegroups.com>
Date: Monday, April 30, 2012 7:45 AM
To: <lrmi@googlegroups.com>
Subject: Re: Some thoughts from our research on curriculum models in
metadata
>For example: I can't imagine a teacher looking for Jane Eyre resources
>typing 'Jane Eyre expositive'.
>Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
>Zoe, it appears that there there is an underlying assumption to this
>thread that metadata utility is all about one aspect of "search": what
>terms get typed into the browser. But there are other critical user
>behaviors with search beyond determining the search string. An equally
>important matter is whether the descriptions returned by whatever is
>typed into the search box contain useful information of a kind sufficient
>to support making a decision whether to retrieve. "P1H30M" or
>"Expositive" may not be terms a teacher initially types into Google, but
>that bit of information might well support a decision to retrieve or not
>when what is needed is a lesson plan that can be usefully deployed in
>under an hour and is purely expositive. This is particularly true when
>result sets are appropriately faceted.
>So, "Jane Eyre lesson plan" may be what is typed as the first step in
>search, but the other structured metadata likely determines whether what
>is returned by the engine is retrieved. So, Zoe, I think you are likely
>spot-on in your analysis if the sole purpose of structured LRMI and
>schema.org <http://schema.org> metadata is the magic bullet of that
>evasive perfect string to type into the search box. But, deciding which
>resources in the retrieved list to actually retrieve is a function beyond
>the search string.
>Stuart
>-- >Stuart A. Sutton, >CEO and Managing Director, Dublin Core Metadata Initiative
>Associate Professor Emeritus, The Information School
>University of Washington
(The post wasn't written for this list and I've chosen not to edit it,
but if you have a read you'll notice that some of my questions about
the nature of usefulness in learning metadata are long-standing.)
(TL;DR: Food is an area where we all agree what constitutes an entity,
and what the names of the entities are. The same applies to movies
(e.g. 'title', 'director'), books ('author', 'publisher'), and so on.
Pretty much every field of human industry *except* learning content,
maybe! In learning, consensus on entity existence, limit, and name is
either loose or not present.)
I do like the google reciple search, very much. I've been using the
yummly.com interface as an example of the same thing in my efforts to
improve our faceted search.
If one goal of LRMI is to support faceted search, however, that
implies the need for the most consensus-driven/discrete/user-
predictable fields we can get - a strong argument, I think, for
considering the inclusion of Discipline and Qualification.
Zoe
On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> Following up on Stuart's response, the model I've had in my head for
> justifying LRMI (and schema.org in general) is the faceted search that
> Google now supports around recipes.
> I'm never sure if this works for everyone, but when I do a search now for
> "potato salad recipe" in Google, I now get a set of facets on the left
> that let me narrow down my search, for ingredients, cook time and calorie
> count.
> As a casual recipe searcher, I think I'd be unlikely to use a
> search term "potato salad recipe with less than 300 calories" much less
> "potato salad recipe with 200 calories." But having that metadata exposed
> to Google allows it to provide a better search experience.
> You can imagine how this might work for LRMI metadata. The user starts
> with a search on "Math," but can quickly see what the range of grade
> levels, type of content, duration, etc are. She can then use a filter to
> narrow the results, or perhaps realize she needs a more specific search
> term.
> I still believe full text searches are where most users want to start, but
> I think there's a lot of room to enhance the search experience with better
> metadata.
> - Peter
> Peter Pinch | Production Manager, OpenCourseWare
> Massachusetts Institute of Technology
> From: Stuart Sutton <stuartasut...@gmail.com>
> Reply-To: <lrmi@googlegroups.com>
> Date: Monday, April 30, 2012 7:45 AM
> To: <lrmi@googlegroups.com>
> Subject: Re: Some thoughts from our research on curriculum models in
> metadata
> >For example: I can't imagine a teacher looking for Jane Eyre resources
> >typing 'Jane Eyre expositive'.
> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
> >Zoe, it appears that there there is an underlying assumption to this
> >thread that metadata utility is all about one aspect of "search": what
> >terms get typed into the browser. But there are other critical user
> >behaviors with search beyond determining the search string. An equally
> >important matter is whether the descriptions returned by whatever is
> >typed into the search box contain useful information of a kind sufficient
> >to support making a decision whether to retrieve. "P1H30M" or
> >"Expositive" may not be terms a teacher initially types into Google, but
> >that bit of information might well support a decision to retrieve or not
> >when what is needed is a lesson plan that can be usefully deployed in
> >under an hour and is purely expositive. This is particularly true when
> >result sets are appropriately faceted.
> >So, "Jane Eyre lesson plan" may be what is typed as the first step in
> >search, but the other structured metadata likely determines whether what
> >is returned by the engine is retrieved. So, Zoe, I think you are likely
> >spot-on in your analysis if the sole purpose of structured LRMI and
> >schema.org <http://schema.org> metadata is the magic bullet of that
> >evasive perfect string to type into the search box. But, deciding which
> >resources in the retrieved list to actually retrieve is a function beyond
> >the search string.
> >Stuart
> >--
> >Stuart A. Sutton,
> >CEO and Managing Director, Dublin Core Metadata Initiative
> >Associate Professor Emeritus, The Information School
> >University of Washington
I just worked out where I think the big difference if for what the
LRMI draft does from what you want it to do (e.g. modelling from the
'food' example...)
The problem is in having full-text outcomes, without contextualising
structures.
So:
'Mayonnaise' will fit on a sidebar.
'Participate in collaborative conversations with diverse partners
about grade 1 topics and texts with peers and adults in small and
larger groups' will not fit on a sidebar.
'Mayonnnaise' exists in the semantic construct 'ingredients'
''Participate in collaborative conversations with diverse partners
about grade 1 topics and texts with peers and adults in small and
larger groups' exists in the semantic construct 'Outcomes'...
...but that's not actually useful information without the
contextualising specifier 'English'. (The outcome could just as easily
be taken from grade 1 'Science'.)
I note that in the string 'Participate in collaborative conversations
with diverse partners about grade 1 topics and texts with peers and
adults in small and larger groups', the string 'English' does not
occur - nor does 'speaking', 'listening', or 'speaking and listening'.
I also note that the LRMI 0.7 draft doesn't have an area where these
meaning-making fields that define the context of the outcome can be
recorded - or where the identifier for the outcome string (which ties
back to the meaning-making fields) can be stored. There's only the
full-text outcome, and as our research has shown, full-text outcomes
almost never contain strings that match user search terms.
My point here is, full-text outcome strings without context don't
support usable display of information pertaining to search results
either.
So while I think your model of 'recipe' is a good one to aim for, I
don't think that the 0.7 specification supports it just yet. That
would require identifiers and subject areas as a minimum, not an
optional extra.
Hope that made sense...
Zoe
On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> Following up on Stuart's response, the model I've had in my head for
> justifying LRMI (and schema.org in general) is the faceted search that
> Google now supports around recipes.
> I'm never sure if this works for everyone, but when I do a search now for
> "potato salad recipe" in Google, I now get a set of facets on the left
> that let me narrow down my search, for ingredients, cook time and calorie
> count.
> As a casual recipe searcher, I think I'd be unlikely to use a
> search term "potato salad recipe with less than 300 calories" much less
> "potato salad recipe with 200 calories." But having that metadata exposed
> to Google allows it to provide a better search experience.
> You can imagine how this might work for LRMI metadata. The user starts
> with a search on "Math," but can quickly see what the range of grade
> levels, type of content, duration, etc are. She can then use a filter to
> narrow the results, or perhaps realize she needs a more specific search
> term.
> I still believe full text searches are where most users want to start, but
> I think there's a lot of room to enhance the search experience with better
> metadata.
> - Peter
> Peter Pinch | Production Manager, OpenCourseWare
> Massachusetts Institute of Technology
> From: Stuart Sutton <stuartasut...@gmail.com>
> Reply-To: <lrmi@googlegroups.com>
> Date: Monday, April 30, 2012 7:45 AM
> To: <lrmi@googlegroups.com>
> Subject: Re: Some thoughts from our research on curriculum models in
> metadata
> >For example: I can't imagine a teacher looking for Jane Eyre resources
> >typing 'Jane Eyre expositive'.
> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
> >Zoe, it appears that there there is an underlying assumption to this
> >thread that metadata utility is all about one aspect of "search": what
> >terms get typed into the browser. But there are other critical user
> >behaviors with search beyond determining the search string. An equally
> >important matter is whether the descriptions returned by whatever is
> >typed into the search box contain useful information of a kind sufficient
> >to support making a decision whether to retrieve. "P1H30M" or
> >"Expositive" may not be terms a teacher initially types into Google, but
> >that bit of information might well support a decision to retrieve or not
> >when what is needed is a lesson plan that can be usefully deployed in
> >under an hour and is purely expositive. This is particularly true when
> >result sets are appropriately faceted.
> >So, "Jane Eyre lesson plan" may be what is typed as the first step in
> >search, but the other structured metadata likely determines whether what
> >is returned by the engine is retrieved. So, Zoe, I think you are likely
> >spot-on in your analysis if the sole purpose of structured LRMI and
> >schema.org <http://schema.org> metadata is the magic bullet of that
> >evasive perfect string to type into the search box. But, deciding which
> >resources in the retrieved list to actually retrieve is a function beyond
> >the search string.
> >Stuart
> >--
> >Stuart A. Sutton,
> >CEO and Managing Director, Dublin Core Metadata Initiative
> >Associate Professor Emeritus, The Information School
> >University of Washington
In the case of relations to competencies, I don't think the text of the
competency is nearly as useful as the metadata around it -- first that a
relation exists at all; then the jurisdiction of the competency; and then,
maybe someday, the position of a competency within a curriculum.
Another thing to keep in mind is one of the basic ideas of microdata --
that the metadata shouldn't take precedence over the data. The primary use
case (at least for schema.org) is of adding semantic markup to an already
existing web page. The presumption is the text that makes the page useful
for humans is already there. The purpose of the microdata is to make it
easier for computers to read.
So you actually have the option of putting something sensible in the text
of the page, perhaps "this resource supports collaboration in grades 1-2"
and then back-it up with microdata that links to the competency.
- Peter
-----------
Peter Pinch
Production Manager, MIT OpenCourseWare
http://ocw.mit.edu
Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
Date: Tuesday, May 1, 2012 8:03 AM
To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
Subject: Re: Some thoughts from our research on curriculum models in
metadata
>Peter -
>I just worked out where I think the big difference if for what the
>LRMI draft does from what you want it to do (e.g. modelling from the
>'food' example...)
>The problem is in having full-text outcomes, without contextualising
>structures.
>So:
>'Mayonnaise' will fit on a sidebar.
>'Participate in collaborative conversations with diverse partners
>about grade 1 topics and texts with peers and adults in small and
>larger groups' will not fit on a sidebar.
>'Mayonnnaise' exists in the semantic construct 'ingredients'
>''Participate in collaborative conversations with diverse partners
>about grade 1 topics and texts with peers and adults in small and
>larger groups' exists in the semantic construct 'Outcomes'...
>...but that's not actually useful information without the
>contextualising specifier 'English'. (The outcome could just as easily
>be taken from grade 1 'Science'.)
>I note that in the string 'Participate in collaborative conversations
>with diverse partners about grade 1 topics and texts with peers and
>adults in small and larger groups', the string 'English' does not
>occur - nor does 'speaking', 'listening', or 'speaking and listening'.
>I also note that the LRMI 0.7 draft doesn't have an area where these
>meaning-making fields that define the context of the outcome can be
>recorded - or where the identifier for the outcome string (which ties
>back to the meaning-making fields) can be stored. There's only the
>full-text outcome, and as our research has shown, full-text outcomes
>almost never contain strings that match user search terms.
>My point here is, full-text outcome strings without context don't
>support usable display of information pertaining to search results
>either.
>So while I think your model of 'recipe' is a good one to aim for, I
>don't think that the 0.7 specification supports it just yet. That
>would require identifiers and subject areas as a minimum, not an
>optional extra.
>Hope that made sense...
>Zoe
>On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
>> Following up on Stuart's response, the model I've had in my head for
>> justifying LRMI (and schema.org in general) is the faceted search that
>> Google now supports around recipes.
>> I'm never sure if this works for everyone, but when I do a search now
>>for
>> "potato salad recipe" in Google, I now get a set of facets on the left
>> that let me narrow down my search, for ingredients, cook time and
>>calorie
>> count.
>> As a casual recipe searcher, I think I'd be unlikely to use a
>> search term "potato salad recipe with less than 300 calories" much less
>> "potato salad recipe with 200 calories." But having that metadata
>>exposed
>> to Google allows it to provide a better search experience.
>> You can imagine how this might work for LRMI metadata. The user starts
>> with a search on "Math," but can quickly see what the range of grade
>> levels, type of content, duration, etc are. She can then use a filter to
>> narrow the results, or perhaps realize she needs a more specific search
>> term.
>> I still believe full text searches are where most users want to start,
>>but
>> I think there's a lot of room to enhance the search experience with
>>better
>> metadata.
>> - Peter
>> Peter Pinch | Production Manager, OpenCourseWare
>> Massachusetts Institute of Technology
>> From: Stuart Sutton <stuartasut...@gmail.com>
>> Reply-To: <lrmi@googlegroups.com>
>> Date: Monday, April 30, 2012 7:45 AM
>> To: <lrmi@googlegroups.com>
>> Subject: Re: Some thoughts from our research on curriculum models in
>> metadata
>> >For example: I can't imagine a teacher looking for Jane Eyre resources
>> >typing 'Jane Eyre expositive'.
>> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
>> >Zoe, it appears that there there is an underlying assumption to this
>> >thread that metadata utility is all about one aspect of "search": what
>> >terms get typed into the browser. But there are other critical user
>> >behaviors with search beyond determining the search string. An equally
>> >important matter is whether the descriptions returned by whatever is
>> >typed into the search box contain useful information of a kind
>>sufficient
>> >to support making a decision whether to retrieve. "P1H30M" or
>> >"Expositive" may not be terms a teacher initially types into Google,
>>but
>> >that bit of information might well support a decision to retrieve or
>>not
>> >when what is needed is a lesson plan that can be usefully deployed in
>> >under an hour and is purely expositive. This is particularly true when
>> >result sets are appropriately faceted.
>> >So, "Jane Eyre lesson plan" may be what is typed as the first step in
>> >search, but the other structured metadata likely determines whether
>>what
>> >is returned by the engine is retrieved. So, Zoe, I think you are
>>likely
>> >spot-on in your analysis if the sole purpose of structured LRMI and
>> >schema.org <http://schema.org> metadata is the magic bullet of that
>> >evasive perfect string to type into the search box. But, deciding
>>which
>> >resources in the retrieved list to actually retrieve is a function
>>beyond
>> >the search string.
>> >Stuart
>> >--
>> >Stuart A. Sutton,
>> >CEO and Managing Director, Dublin Core Metadata Initiative
>> >Associate Professor Emeritus, The Information School
>> >University of Washington
So, Peter, what do you recommend? Leaving things as they are, or
modifying them to reflect the discussion?
(On the point of pre-existing html - that's what I thought the
intention was, but I wasn't sure as 'grade' isn't currently a field,
only age range is. Yet very little of our learning content (and the
learning content I see around the web) is marked up with ages, it's
marked up with grades or exam levels. So we wouldn't be able to use
the 'age' field.
Additional to this - outcomes/competencies are not objectively
appropriate for any given age, they are only meaningful in the context
of the whole curriculum. Assuming that Texas' outcomes for eight year
olds are the same as France's outcomes for eight year olds is asking
for trouble - another reason why I'm starting to suspect the outcome
context (grade, subject area, qualification) are required rather than
optional for LRMI metadata to be meaningful.
Zoe
On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> In the case of relations to competencies, I don't think the text of the
> competency is nearly as useful as the metadata around it -- first that a
> relation exists at all; then the jurisdiction of the competency; and then,
> maybe someday, the position of a competency within a curriculum.
> Another thing to keep in mind is one of the basic ideas of microdata --
> that the metadata shouldn't take precedence over the data. The primary use
> case (at least for schema.org) is of adding semantic markup to an already
> existing web page. The presumption is the text that makes the page useful
> for humans is already there. The purpose of the microdata is to make it
> easier for computers to read.
> So you actually have the option of putting something sensible in the text
> of the page, perhaps "this resource supports collaboration in grades 1-2"
> and then back-it up with microdata that links to the competency.
> - Peter
> -----------
> Peter Pinch
> Production Manager, MIT OpenCourseWarehttp://ocw.mit.edu
> Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
> Date: Tuesday, May 1, 2012 8:03 AM
> To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
> Subject: Re: Some thoughts from our research on curriculum models in
> metadata
> >Peter -
> >I just worked out where I think the big difference if for what the
> >LRMI draft does from what you want it to do (e.g. modelling from the
> >'food' example...)
> >The problem is in having full-text outcomes, without contextualising
> >structures.
> >So:
> >'Mayonnaise' will fit on a sidebar.
> >'Participate in collaborative conversations with diverse partners
> >about grade 1 topics and texts with peers and adults in small and
> >larger groups' will not fit on a sidebar.
> >'Mayonnnaise' exists in the semantic construct 'ingredients'
> >''Participate in collaborative conversations with diverse partners
> >about grade 1 topics and texts with peers and adults in small and
> >larger groups' exists in the semantic construct 'Outcomes'...
> >...but that's not actually useful information without the
> >contextualising specifier 'English'. (The outcome could just as easily
> >be taken from grade 1 'Science'.)
> >I note that in the string 'Participate in collaborative conversations
> >with diverse partners about grade 1 topics and texts with peers and
> >adults in small and larger groups', the string 'English' does not
> >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> >I also note that the LRMI 0.7 draft doesn't have an area where these
> >meaning-making fields that define the context of the outcome can be
> >recorded - or where the identifier for the outcome string (which ties
> >back to the meaning-making fields) can be stored. There's only the
> >full-text outcome, and as our research has shown, full-text outcomes
> >almost never contain strings that match user search terms.
> >My point here is, full-text outcome strings without context don't
> >support usable display of information pertaining to search results
> >either.
> >So while I think your model of 'recipe' is a good one to aim for, I
> >don't think that the 0.7 specification supports it just yet. That
> >would require identifiers and subject areas as a minimum, not an
> >optional extra.
> >Hope that made sense...
> >Zoe
> >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> >> Following up on Stuart's response, the model I've had in my head for
> >> justifying LRMI (and schema.org in general) is the faceted search that
> >> Google now supports around recipes.
> >> I'm never sure if this works for everyone, but when I do a search now
> >>for
> >> "potato salad recipe" in Google, I now get a set of facets on the left
> >> that let me narrow down my search, for ingredients, cook time and
> >>calorie
> >> count.
> >> As a casual recipe searcher, I think I'd be unlikely to use a
> >> search term "potato salad recipe with less than 300 calories" much less
> >> "potato salad recipe with 200 calories." But having that metadata
> >>exposed
> >> to Google allows it to provide a better search experience.
> >> You can imagine how this might work for LRMI metadata. The user starts
> >> with a search on "Math," but can quickly see what the range of grade
> >> levels, type of content, duration, etc are. She can then use a filter to
> >> narrow the results, or perhaps realize she needs a more specific search
> >> term.
> >> I still believe full text searches are where most users want to start,
> >>but
> >> I think there's a lot of room to enhance the search experience with
> >>better
> >> metadata.
> >> - Peter
> >> Peter Pinch | Production Manager, OpenCourseWare
> >> Massachusetts Institute of Technology
> >> From: Stuart Sutton <stuartasut...@gmail.com>
> >> Reply-To: <lrmi@googlegroups.com>
> >> Date: Monday, April 30, 2012 7:45 AM
> >> To: <lrmi@googlegroups.com>
> >> Subject: Re: Some thoughts from our research on curriculum models in
> >> metadata
> >> >For example: I can't imagine a teacher looking for Jane Eyre resources
> >> >typing 'Jane Eyre expositive'.
> >> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
> >> >Zoe, it appears that there there is an underlying assumption to this
> >> >thread that metadata utility is all about one aspect of "search": what
> >> >terms get typed into the browser. But there are other critical user
> >> >behaviors with search beyond determining the search string. An equally
> >> >important matter is whether the descriptions returned by whatever is
> >> >typed into the search box contain useful information of a kind
> >>sufficient
> >> >to support making a decision whether to retrieve. "P1H30M" or
> >> >"Expositive" may not be terms a teacher initially types into Google,
> >>but
> >> >that bit of information might well support a decision to retrieve or
> >>not
> >> >when what is needed is a lesson plan that can be usefully deployed in
> >> >under an hour and is purely expositive. This is particularly true when
> >> >result sets are appropriately faceted.
> >> >So, "Jane Eyre lesson plan" may be what is typed as the first step in
> >> >search, but the other structured metadata likely determines whether
> >>what
> >> >is returned by the engine is retrieved. So, Zoe, I think you are
> >>likely
> >> >spot-on in your analysis if the sole purpose of structured LRMI and
> >> >schema.org <http://schema.org> metadata is the magic bullet of that
> >> >evasive perfect string to type into the search box. But, deciding
> >>which
> >> >resources in the retrieved list to actually retrieve is a function
> >>beyond
> >> >the search string.
> >> >Stuart
> >> >--
> >> >Stuart A. Sutton,
> >> >CEO and Managing Director, Dublin Core Metadata Initiative
> >> >Associate Professor Emeritus, The Information School
> >> >University of Washington
Zoe, dcterms:educationLevel [1] used by many Dublin Core-based projects for
expressing the notion behind grades (in U.S. vernacular) and qualifications
was part of the original listing of potential LRMI properties but was not
included. I have no idea whether it is possible at this point to reopen
those discussions/decisions since finalization of the LRMI specification is
near.
However, you are correct, the prevalent expressions (particularly among
teachers/searchers) are in terms of grades/qualifications and not typical
age range. I guess this assumes many projects will be mapping from
education levels to typical age [see 2] or using some other non-LRMI
property.
On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com> wrote:
> So, Peter, what do you recommend? Leaving things as they are, or
> modifying them to reflect the discussion?
> (On the point of pre-existing html - that's what I thought the
> intention was, but I wasn't sure as 'grade' isn't currently a field,
> only age range is. Yet very little of our learning content (and the
> learning content I see around the web) is marked up with ages, it's
> marked up with grades or exam levels. So we wouldn't be able to use
> the 'age' field.
> Additional to this - outcomes/competencies are not objectively
> appropriate for any given age, they are only meaningful in the context
> of the whole curriculum. Assuming that Texas' outcomes for eight year
> olds are the same as France's outcomes for eight year olds is asking
> for trouble - another reason why I'm starting to suspect the outcome
> context (grade, subject area, qualification) are required rather than
> optional for LRMI metadata to be meaningful.
> Zoe
> On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > Absolutely makes sense to me.
> > In the case of relations to competencies, I don't think the text of the
> > competency is nearly as useful as the metadata around it -- first that a
> > relation exists at all; then the jurisdiction of the competency; and
> then,
> > maybe someday, the position of a competency within a curriculum.
> > Another thing to keep in mind is one of the basic ideas of microdata --
> > that the metadata shouldn't take precedence over the data. The primary
> use
> > case (at least for schema.org) is of adding semantic markup to an
> already
> > existing web page. The presumption is the text that makes the page useful
> > for humans is already there. The purpose of the microdata is to make it
> > easier for computers to read.
> > So you actually have the option of putting something sensible in the text
> > of the page, perhaps "this resource supports collaboration in grades 1-2"
> > and then back-it up with microdata that links to the competency.
> > - Peter
> > -----------
> > Peter Pinch
> > Production Manager, MIT OpenCourseWarehttp://ocw.mit.edu
> > Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
> > Date: Tuesday, May 1, 2012 8:03 AM
> > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
> > Subject: Re: Some thoughts from our research on curriculum models in
> > metadata
> > >Peter -
> > >I just worked out where I think the big difference if for what the
> > >LRMI draft does from what you want it to do (e.g. modelling from the
> > >'food' example...)
> > >The problem is in having full-text outcomes, without contextualising
> > >structures.
> > >So:
> > >'Mayonnaise' will fit on a sidebar.
> > >'Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' will not fit on a sidebar.
> > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > >''Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' exists in the semantic construct 'Outcomes'...
> > >...but that's not actually useful information without the
> > >contextualising specifier 'English'. (The outcome could just as easily
> > >be taken from grade 1 'Science'.)
> > >I note that in the string 'Participate in collaborative conversations
> > >with diverse partners about grade 1 topics and texts with peers and
> > >adults in small and larger groups', the string 'English' does not
> > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > >meaning-making fields that define the context of the outcome can be
> > >recorded - or where the identifier for the outcome string (which ties
> > >back to the meaning-making fields) can be stored. There's only the
> > >full-text outcome, and as our research has shown, full-text outcomes
> > >almost never contain strings that match user search terms.
> > >My point here is, full-text outcome strings without context don't
> > >support usable display of information pertaining to search results
> > >either.
> > >So while I think your model of 'recipe' is a good one to aim for, I
> > >don't think that the 0.7 specification supports it just yet. That
> > >would require identifiers and subject areas as a minimum, not an
> > >optional extra.
> > >Hope that made sense...
> > >Zoe
> > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > >> Following up on Stuart's response, the model I've had in my head for
> > >> justifying LRMI (and schema.org in general) is the faceted search
> that
> > >> Google now supports around recipes.
> > >> I'm never sure if this works for everyone, but when I do a search now
> > >>for
> > >> "potato salad recipe" in Google, I now get a set of facets on the left
> > >> that let me narrow down my search, for ingredients, cook time and
> > >>calorie
> > >> count.
> > >> As a casual recipe searcher, I think I'd be unlikely to use a
> > >> search term "potato salad recipe with less than 300 calories" much
> less
> > >> "potato salad recipe with 200 calories." But having that metadata
> > >>exposed
> > >> to Google allows it to provide a better search experience.
> > >> You can imagine how this might work for LRMI metadata. The user starts
> > >> with a search on "Math," but can quickly see what the range of grade
> > >> levels, type of content, duration, etc are. She can then use a filter
> to
> > >> narrow the results, or perhaps realize she needs a more specific
> search
> > >> term.
> > >> I still believe full text searches are where most users want to start,
> > >>but
> > >> I think there's a lot of room to enhance the search experience with
> > >>better
> > >> metadata.
> > >> - Peter
> > >> Peter Pinch | Production Manager, OpenCourseWare
> > >> Massachusetts Institute of Technology
> > >> From: Stuart Sutton <stuartasut...@gmail.com>
> > >> Reply-To: <lrmi@googlegroups.com>
> > >> Date: Monday, April 30, 2012 7:45 AM
> > >> To: <lrmi@googlegroups.com>
> > >> Subject: Re: Some thoughts from our research on curriculum models in
> > >> metadata
> > >> >Here are some terms that I can't imagine an educational-resource-
> > >> >seeking user typing into google, either with or without other
> strings:
> > >> >For example: I can't imagine a teacher looking for Jane Eyre
> resources
> > >> >typing 'Jane Eyre expositive'.
> > >> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
> > >> >Zoe, it appears that there there is an underlying assumption to this
> > >> >thread that metadata utility is all about one aspect of "search":
> what
> > >> >terms get typed into the browser. But there are other critical user
> > >> >behaviors with search beyond determining the search string. An
> equally
> > >> >important matter is whether the descriptions returned by whatever is
> > >> >typed into the search box contain useful information of a kind
> > >>sufficient
> > >> >to support making a decision whether to retrieve. "P1H30M" or
> > >> >"Expositive" may not be terms a teacher initially types into Google,
> > >>but
> > >> >that bit of information might well support a decision to retrieve or
> > >>not
> > >> >when what is needed is a lesson plan that can be usefully deployed in
> > >> >under an hour and is purely expositive. This is particularly true
> when
> > >> >result sets are appropriately faceted.
> > >> >So, "Jane Eyre lesson plan" may be what is typed as the first step in
> > >> >search, but the other structured metadata likely determines whether
> > >>what
> > >> >is returned by the engine is retrieved. So, Zoe, I think you are
> > >>likely
> > >> >spot-on in your analysis if the sole purpose of structured LRMI and
> > >> >schema.org <http://schema.org> metadata is the magic bullet of that
> > >> >evasive perfect string to type into the search box. But, deciding
> > >>which
> > >> >resources in the retrieved list to actually retrieve is a function
> > >>beyond
> > >> >the search string.
Yes you are correct, we discussed mapping typical age to ed level. We also said this would not work for higher education and then sort of left it there. However, since we have the box of worm cans open, we might as well look at this educaitonalLevel issues again (Which seems like yet another framework alignment issue… e.g. What is your ed level framework and which level is this resource aligned with.)
Joshua Marks
CTO
Curriki: The Global Education and Learning Community
From: lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] On Behalf Of Stuart Sutton
Sent: Tuesday, May 01, 2012 8:54 AM
To: lrmi@googlegroups.com
Subject: Re: Some thoughts from our research on curriculum models in metadata
Zoe, dcterms:educationLevel [1] used by many Dublin Core-based projects for expressing the notion behind grades (in U.S. vernacular) and qualifications was part of the original listing of potential LRMI properties but was not included. I have no idea whether it is possible at this point to reopen those discussions/decisions since finalization of the LRMI specification is near.
However, you are correct, the prevalent expressions (particularly among teachers/searchers) are in terms of grades/qualifications and not typical age range. I guess this assumes many projects will be mapping from education levels to typical age [see 2] or using some other non-LRMI property.
On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com> wrote:
So, Peter, what do you recommend? Leaving things as they are, or
modifying them to reflect the discussion?
(On the point of pre-existing html - that's what I thought the
intention was, but I wasn't sure as 'grade' isn't currently a field,
only age range is. Yet very little of our learning content (and the
learning content I see around the web) is marked up with ages, it's
marked up with grades or exam levels. So we wouldn't be able to use
the 'age' field.
Additional to this - outcomes/competencies are not objectively
appropriate for any given age, they are only meaningful in the context
of the whole curriculum. Assuming that Texas' outcomes for eight year
olds are the same as France's outcomes for eight year olds is asking
for trouble - another reason why I'm starting to suspect the outcome
context (grade, subject area, qualification) are required rather than
optional for LRMI metadata to be meaningful.
Zoe
On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> In the case of relations to competencies, I don't think the text of the
> competency is nearly as useful as the metadata around it -- first that a
> relation exists at all; then the jurisdiction of the competency; and then,
> maybe someday, the position of a competency within a curriculum.
> Another thing to keep in mind is one of the basic ideas of microdata --
> that the metadata shouldn't take precedence over the data. The primary use
> case (at least for schema.org) is of adding semantic markup to an already
> existing web page. The presumption is the text that makes the page useful
> for humans is already there. The purpose of the microdata is to make it
> easier for computers to read.
> So you actually have the option of putting something sensible in the text
> of the page, perhaps "this resource supports collaboration in grades 1-2"
> and then back-it up with microdata that links to the competency.
> - Peter
> -----------
> Peter Pinch
> Production Manager, MIT OpenCourseWarehttp://ocw.mit.edu
> Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
> Date: Tuesday, May 1, 2012 8:03 AM
> To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
> Subject: Re: Some thoughts from our research on curriculum models in
> metadata
> >Peter -
> >I just worked out where I think the big difference if for what the
> >LRMI draft does from what you want it to do (e.g. modelling from the
> >'food' example...)
> >The problem is in having full-text outcomes, without contextualising
> >structures.
> >So:
> >'Mayonnaise' will fit on a sidebar.
> >'Participate in collaborative conversations with diverse partners
> >about grade 1 topics and texts with peers and adults in small and
> >larger groups' will not fit on a sidebar.
> >'Mayonnnaise' exists in the semantic construct 'ingredients'
> >''Participate in collaborative conversations with diverse partners
> >about grade 1 topics and texts with peers and adults in small and
> >larger groups' exists in the semantic construct 'Outcomes'...
> >...but that's not actually useful information without the
> >contextualising specifier 'English'. (The outcome could just as easily
> >be taken from grade 1 'Science'.)
> >I note that in the string 'Participate in collaborative conversations
> >with diverse partners about grade 1 topics and texts with peers and
> >adults in small and larger groups', the string 'English' does not
> >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> >I also note that the LRMI 0.7 draft doesn't have an area where these
> >meaning-making fields that define the context of the outcome can be
> >recorded - or where the identifier for the outcome string (which ties
> >back to the meaning-making fields) can be stored. There's only the
> >full-text outcome, and as our research has shown, full-text outcomes
> >almost never contain strings that match user search terms.
> >My point here is, full-text outcome strings without context don't
> >support usable display of information pertaining to search results
> >either.
> >So while I think your model of 'recipe' is a good one to aim for, I
> >don't think that the 0.7 specification supports it just yet. That
> >would require identifiers and subject areas as a minimum, not an
> >optional extra.
> >Hope that made sense...
> >Zoe
> >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> >> Following up on Stuart's response, the model I've had in my head for
> >> justifying LRMI (and schema.org in general) is the faceted search that
> >> Google now supports around recipes.
> >> I'm never sure if this works for everyone, but when I do a search now
> >>for
> >> "potato salad recipe" in Google, I now get a set of facets on the left
> >> that let me narrow down my search, for ingredients, cook time and
> >>calorie
> >> count.
> >> As a casual recipe searcher, I think I'd be unlikely to use a
> >> search term "potato salad recipe with less than 300 calories" much less
> >> "potato salad recipe with 200 calories." But having that metadata
> >>exposed
> >> to Google allows it to provide a better search experience.
> >> You can imagine how this might work for LRMI metadata. The user starts
> >> with a search on "Math," but can quickly see what the range of grade
> >> levels, type of content, duration, etc are. She can then use a filter to
> >> narrow the results, or perhaps realize she needs a more specific search
> >> term.
> >> I still believe full text searches are where most users want to start,
> >>but
> >> I think there's a lot of room to enhance the search experience with
> >>better
> >> metadata.
> >> - Peter
> >> Peter Pinch | Production Manager, OpenCourseWare
> >> Massachusetts Institute of Technology
> >> From: Stuart Sutton <stuartasut...@gmail.com>
> >> Reply-To: <lrmi@googlegroups.com>
> >> Date: Monday, April 30, 2012 7:45 AM
> >> To: <lrmi@googlegroups.com>
> >> Subject: Re: Some thoughts from our research on curriculum models in
> >> metadata
> >> >For example: I can't imagine a teacher looking for Jane Eyre resources
> >> >typing 'Jane Eyre expositive'.
> >> >Nor can I imagine the string 'Jane Eyre assesses' or 'Jane Eyre P1H'.
> >> >Zoe, it appears that there there is an underlying assumption to this
> >> >thread that metadata utility is all about one aspect of "search": what
> >> >terms get typed into the browser. But there are other critical user
> >> >behaviors with search beyond determining the search string. An equally
> >> >important matter is whether the descriptions returned by whatever is
> >> >typed into the search box contain useful information of a kind
> >>sufficient
> >> >to support making a decision whether to retrieve. "P1H30M" or
> >> >"Expositive" may not be terms a teacher initially types into Google,
> >>but
> >> >that bit of information might well support a decision to retrieve or
> >>not
> >> >when what is needed is a lesson plan that can be usefully deployed in
> >> >under an hour and is purely expositive. This is particularly true when
> >> >result
Joshua, I am not certain that for most cases an assertion of educationLevel
need come even close to the potential complexity of an assertion of a
correlation between a leaning resource and authoritatively promulgated
learning outcome. While I will not venture to say anything about other
jurisdictions, there are longstanding definitions of education levels in
the US. [1] was developed in the mid-1990s for what was then the U.S.
Department of Education's Gateway to Educational Materials (GEM) [now the
Gateway for 21st Century Skills]--also later published for use with NSDL.
So, the means for making straightforward assertions exist. Also
http://en.wikipedia.org/wiki/Educational_stages demonstrates that there are
fairly well defined frameworks in most jurisdictions. Mapping a learning
resource to particular points in such frameworks is no harder (perhaps
less) than mapping to a typical age range.
On Tue, May 1, 2012 at 9:06 AM, Joshua Marks <jma...@curriki.org> wrote:
> Stuart,****
> ** **
> Yes you are correct, we discussed mapping typical age to ed level. We also
> said this would not work for higher education and then sort of left it
> there. However, since we have the box of worm cans open, we might as well
> look at this educaitonalLevel issues again (Which seems like yet another
> framework alignment issue… e.g. What is your ed level framework and which
> level is this resource aligned with.) ****
> ** **
> Joshua Marks****
> CTO****
> Curriki: The Global Education and Learning Community ****
> *From:* lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> Of *Stuart Sutton
> *Sent:* Tuesday, May 01, 2012 8:54 AM
> *To:* lrmi@googlegroups.com
> *Subject:* Re: Some thoughts from our research on curriculum models in
> metadata****
> ** **
> Zoe, dcterms:educationLevel [1] used by many Dublin Core-based projects
> for expressing the notion behind grades (in U.S. vernacular) and
> qualifications was part of the original listing of potential LRMI
> properties but was not included. I have no idea whether it is possible at
> this point to reopen those discussions/decisions since finalization of the
> LRMI specification is near.
> However, you are correct, the prevalent expressions (particularly among
> teachers/searchers) are in terms of grades/qualifications and not typical
> age range. I guess this assumes many projects will be mapping from
> education levels to typical age [see 2] or using some other non-LRMI
> property.
> On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com> wrote:****
> So, Peter, what do you recommend? Leaving things as they are, or
> modifying them to reflect the discussion?
> (On the point of pre-existing html - that's what I thought the
> intention was, but I wasn't sure as 'grade' isn't currently a field,
> only age range is. Yet very little of our learning content (and the
> learning content I see around the web) is marked up with ages, it's
> marked up with grades or exam levels. So we wouldn't be able to use
> the 'age' field.
> Additional to this - outcomes/competencies are not objectively
> appropriate for any given age, they are only meaningful in the context
> of the whole curriculum. Assuming that Texas' outcomes for eight year
> olds are the same as France's outcomes for eight year olds is asking
> for trouble - another reason why I'm starting to suspect the outcome
> context (grade, subject area, qualification) are required rather than
> optional for LRMI metadata to be meaningful.
> Zoe****
> On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > Absolutely makes sense to me.
> > In the case of relations to competencies, I don't think the text of the
> > competency is nearly as useful as the metadata around it -- first that a
> > relation exists at all; then the jurisdiction of the competency; and
> then,
> > maybe someday, the position of a competency within a curriculum.
> > Another thing to keep in mind is one of the basic ideas of microdata --
> > that the metadata shouldn't take precedence over the data. The primary
> use
> > case (at least for schema.org) is of adding semantic markup to an
> already
> > existing web page. The presumption is the text that makes the page useful
> > for humans is already there. The purpose of the microdata is to make it
> > easier for computers to read.
> > So you actually have the option of putting something sensible in the text
> > of the page, perhaps "this resource supports collaboration in grades 1-2"
> > and then back-it up with microdata that links to the competency.
> > Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
> > Date: Tuesday, May 1, 2012 8:03 AM
> > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
> > Subject: Re: Some thoughts from our research on curriculum models in
> > metadata
> > >Peter -
> > >I just worked out where I think the big difference if for what the
> > >LRMI draft does from what you want it to do (e.g. modelling from the
> > >'food' example...)
> > >The problem is in having full-text outcomes, without contextualising
> > >structures.
> > >So:
> > >'Mayonnaise' will fit on a sidebar.
> > >'Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' will not fit on a sidebar.
> > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > >''Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' exists in the semantic construct 'Outcomes'...
> > >...but that's not actually useful information without the
> > >contextualising specifier 'English'. (The outcome could just as easily
> > >be taken from grade 1 'Science'.)
> > >I note that in the string 'Participate in collaborative conversations
> > >with diverse partners about grade 1 topics and texts with peers and
> > >adults in small and larger groups', the string 'English' does not
> > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > >meaning-making fields that define the context of the outcome can be
> > >recorded - or where the identifier for the outcome string (which ties
> > >back to the meaning-making fields) can be stored. There's only the
> > >full-text outcome, and as our research has shown, full-text outcomes
> > >almost never contain strings that match user search terms.
> > >My point here is, full-text outcome strings without context don't
> > >support usable display of information pertaining to search results
> > >either.
> > >So while I think your model of 'recipe' is a good one to aim for, I
> > >don't think that the 0.7 specification supports it just yet. That
> > >would require identifiers and subject areas as a minimum, not an
> > >optional extra.
> > >Hope that made sense...
> > >Zoe
> > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > >> Following up on Stuart's response, the model I've had in my head for
> > >> justifying LRMI (and schema.org in general) is the faceted search
> that
> > >> Google now supports around recipes.
> > >> I'm never sure if this works for everyone, but when I do a search now
> > >>for
> > >> "potato salad recipe" in Google, I now get a set of facets on the left
> > >> that let me narrow down my search, for ingredients, cook time and
> > >>calorie
> > >> count.
> > >> As a casual recipe searcher, I think I'd be unlikely to use a
> > >> search term "potato salad recipe with less than 300 calories" much
> less
> > >> "potato salad recipe with 200 calories." But having that metadata
> > >>exposed
> > >> to Google allows it to provide a better search experience.
> > >> You can imagine how this might work for LRMI metadata. The user starts
> > >> with a search on "Math," but can quickly see what the range of grade
> > >> levels, type of content, duration, etc are. She can then use a filter
> to
> > >> narrow the results, or perhaps realize she needs a more specific
> search
> > >> term.
> > >> I still believe full text searches are where most users want to start,
> > >>but
> > >> I think there's a lot of room to enhance the search experience with
> > >>better
> > >> metadata.
> > >> - Peter
> > >> Peter Pinch | Production Manager, OpenCourseWare
> > >> Massachusetts Institute of Technology
Age doesn't work at all well for HE either, unless you just say 18+ and don't differentiate between 1st year bachelor level and Master degree level.
I think there was also an issue that grade levels are very difficult to map across different education systems.
Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th Grade "usually the first year of high school" equates to Grade 8 for some parts of Australia and Grade 7 in others, if you map via it being "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
Year 10 in England and Wales but Year 11 in Northern Ireland, which for some is the final year of High School not the first...
At which point you perhaps think that maybe the only way to express this equivalence, albeit inadequately, is in terms of the typical age of the students? E.g.
This resource is suitable for Scottish <span itemprop="typicalAge" value="14-15">S5</span> pupils
> Yes you are correct, we discussed mapping typical age to ed level. We > also said this would not work for higher education and then sort of > left it there. However, since we have the box of worm cans open, we > might as well look at this educaitonalLevel issues again (Which seems > like yet another framework alignment issue… e.g. What is your ed level > framework and which level is this resource aligned with.)
> Joshua Marks
> CTO
> Curriki: The Global Education and Learning Community
> *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf > Of *Stuart Sutton
> *Sent:* Tuesday, May 01, 2012 8:54 AM
> *To:* lrmi@googlegroups.com
> *Subject:* Re: Some thoughts from our research on curriculum models in > metadata
> Zoe, dcterms:educationLevel [1] used by many Dublin Core-based > projects for expressing the notion behind grades (in U.S. vernacular) > and qualifications was part of the original listing of potential LRMI > properties but was not included. I have no idea whether it is > possible at this point to reopen those discussions/decisions since > finalization of the LRMI specification is near.
> However, you are correct, the prevalent expressions (particularly > among teachers/searchers) are in terms of grades/qualifications and > not typical age range. I guess this assumes many projects will be > mapping from education levels to typical age [see 2] or using some > other non-LRMI property.
> On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com > <mailto:zoe.f.r...@gmail.com>> wrote:
> So, Peter, what do you recommend? Leaving things as they are, or
> modifying them to reflect the discussion?
> (On the point of pre-existing html - that's what I thought the
> intention was, but I wasn't sure as 'grade' isn't currently a field,
> only age range is. Yet very little of our learning content (and the
> learning content I see around the web) is marked up with ages, it's
> marked up with grades or exam levels. So we wouldn't be able to use
> the 'age' field.
> Additional to this - outcomes/competencies are not objectively
> appropriate for any given age, they are only meaningful in the context
> of the whole curriculum. Assuming that Texas' outcomes for eight year
> olds are the same as France's outcomes for eight year olds is asking
> for trouble - another reason why I'm starting to suspect the outcome
> context (grade, subject area, qualification) are required rather than
> optional for LRMI metadata to be meaningful.
> Zoe
> On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU > <mailto:pdpi...@MIT.EDU>> wrote:
> > Absolutely makes sense to me.
> > In the case of relations to competencies, I don't think the text of the
> > competency is nearly as useful as the metadata around it -- first that a
> > relation exists at all; then the jurisdiction of the competency; and > then,
> > maybe someday, the position of a competency within a curriculum.
> > Another thing to keep in mind is one of the basic ideas of microdata --
> > that the metadata shouldn't take precedence over the data. The > primary use
> > case (at least for schema.org <http://schema.org>) is of adding > semantic markup to an already
> > existing web page. The presumption is the text that makes the page > useful
> > for humans is already there. The purpose of the microdata is to make it
> > easier for computers to read.
> > So you actually have the option of putting something sensible in the > text
> > of the page, perhaps "this resource supports collaboration in grades > 1-2"
> > and then back-it up with microdata that links to the competency.
> > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>" > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > Date: Tuesday, May 1, 2012 8:03 AM
> > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com > <mailto:lrmi@googlegroups.com>>
> > Subject: Re: Some thoughts from our research on curriculum models in
> > metadata
> > >Peter -
> > >I just worked out where I think the big difference if for what the
> > >LRMI draft does from what you want it to do (e.g. modelling from the
> > >'food' example...)
> > >The problem is in having full-text outcomes, without contextualising
> > >structures.
> > >So:
> > >'Mayonnaise' will fit on a sidebar.
> > >'Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' will not fit on a sidebar.
> > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > >''Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' exists in the semantic construct 'Outcomes'...
> > >...but that's not actually useful information without the
> > >contextualising specifier 'English'. (The outcome could just as easily
> > >be taken from grade 1 'Science'.)
> > >I note that in the string 'Participate in collaborative conversations
> > >with diverse partners about grade 1 topics and texts with peers and
> > >adults in small and larger groups', the string 'English' does not
> > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > >meaning-making fields that define the context of the outcome can be
> > >recorded - or where the identifier for the outcome string (which ties
> > >back to the meaning-making fields) can be stored. There's only the
> > >full-text outcome, and as our research has shown, full-text outcomes
> > >almost never contain strings that match user search terms.
> > >My point here is, full-text outcome strings without context don't
> > >support usable display of information pertaining to search results
> > >either.
> > >So while I think your model of 'recipe' is a good one to aim for, I
> > >don't think that the 0.7 specification supports it just yet. That
> > >would require identifiers and subject areas as a minimum, not an
> > >optional extra.
> > >Hope that made sense...
> > >Zoe
> > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU > <mailto:pdpi...@MIT.EDU>> wrote:
> > >> Following up on Stuart's response, the model I've had in my head for
> > >> justifying LRMI (and schema.org <http://schema.org> in general) > is the faceted search that
> > >> Google now supports around recipes.
> > >> I'm never sure if this works for everyone, but when I do a search now
> > >>for
> > >> "potato salad recipe" in Google, I now get a set of facets on the > left
> > >> that let me narrow down my search, for ingredients, cook time and
> > >>calorie
> > >> count.
> > >> As a casual recipe searcher, I think I'd be unlikely to use a
> > >> search term "potato salad recipe with less than 300 calories" > much less
> > >> "potato salad recipe with 200 calories." But having that metadata
> > >>exposed
> > >> to Google allows it to provide a better search experience.
> > >> You can imagine how this might work for LRMI metadata. The user > starts
> > >> with a search on "Math," but can quickly see what the range of grade
> > >> levels, type of content, duration, etc are. She can then use a > filter to
> > >> narrow the results, or perhaps realize she needs a more specific > search
> > >> term.
> > >> I still believe full text searches are where most users want to > start,
> > >>but
> > >> I think there's a lot of room
> Age doesn't work at all well for HE either, unless you just say 18+ and
> don't differentiate between 1st year bachelor level and Master degree level.
> I think there was also an issue that grade levels are very difficult to
> map across different education systems.
> Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> Grade "usually the first year of high school" equates to Grade 8 for
> some parts of Australia and Grade 7 in others, if you map via it being
> "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
> Year 10 in England and Wales but Year 11 in Northern Ireland, which for
> some is the final year of High School not the first...
> At which point you perhaps think that maybe the only way to express this
> equivalence, albeit inadequately, is in terms of the typical age of the
> students? E.g.
> This resource is suitable for Scottish <span itemprop="typicalAge"
> value="14-15">S5</span> pupils
> > Yes you are correct, we discussed mapping typical age to ed level. We
> > also said this would not work for higher education and then sort of
> > left it there. However, since we have the box of worm cans open, we
> > might as well look at this educaitonalLevel issues again (Which seems
> > like yet another framework alignment issue… e.g. What is your ed level
> > framework and which level is this resource aligned with.)
> > Joshua Marks
> > CTO
> > Curriki: The Global Education and Learning Community
> > *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> > Of *Stuart Sutton
> > *Sent:* Tuesday, May 01, 2012 8:54 AM
> > *To:* lrmi@googlegroups.com
> > *Subject:* Re: Some thoughts from our research on curriculum models in
> > metadata
> > Zoe, dcterms:educationLevel [1] used by many Dublin Core-based
> > projects for expressing the notion behind grades (in U.S. vernacular)
> > and qualifications was part of the original listing of potential LRMI
> > properties but was not included. I have no idea whether it is
> > possible at this point to reopen those discussions/decisions since
> > finalization of the LRMI specification is near.
> > However, you are correct, the prevalent expressions (particularly
> > among teachers/searchers) are in terms of grades/qualifications and
> > not typical age range. I guess this assumes many projects will be
> > mapping from education levels to typical age [see 2] or using some
> > other non-LRMI property.
> > On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com
> > <mailto:zoe.f.r...@gmail.com>> wrote:
> > So, Peter, what do you recommend? Leaving things as they are, or
> > modifying them to reflect the discussion?
> > (On the point of pre-existing html - that's what I thought the
> > intention was, but I wasn't sure as 'grade' isn't currently a field,
> > only age range is. Yet very little of our learning content (and the
> > learning content I see around the web) is marked up with ages, it's
> > marked up with grades or exam levels. So we wouldn't be able to use
> > the 'age' field.
> > Additional to this - outcomes/competencies are not objectively
> > appropriate for any given age, they are only meaningful in the context
> > of the whole curriculum. Assuming that Texas' outcomes for eight year
> > olds are the same as France's outcomes for eight year olds is asking
> > for trouble - another reason why I'm starting to suspect the outcome
> > context (grade, subject area, qualification) are required rather than
> > optional for LRMI metadata to be meaningful.
> > Zoe
> > On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU
> > <mailto:pdpi...@MIT.EDU>> wrote:
> > > Absolutely makes sense to me.
> > > In the case of relations to competencies, I don't think the text of the
> > > competency is nearly as useful as the metadata around it -- first that a
> > > relation exists at all; then the jurisdiction of the competency; and
> > then,
> > > maybe someday, the position of a competency within a curriculum.
> > > Another thing to keep in mind is one of the basic ideas of microdata --
> > > that the metadata shouldn't take precedence over the data. The
> > primary use
> > > case (at least for schema.org <http://schema.org>) is of adding
> > semantic markup to an already
> > > existing web page. The presumption is the text that makes the page
> > useful
> > > for humans is already there. The purpose of the microdata is to make it
> > > easier for computers to read.
> > > So you actually have the option of putting something sensible in the
> > text
> > > of the page, perhaps "this resource supports collaboration in grades
> > 1-2"
> > > and then back-it up with microdata that links to the competency.
> > > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>"
> > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > > Date: Tuesday, May 1, 2012 8:03 AM
> > > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com
> > <mailto:lrmi@googlegroups.com>>
> > > Subject: Re: Some thoughts from our research on curriculum models in
> > > metadata
> > > >Peter -
> > > >I just worked out where I think the big difference if for what the
> > > >LRMI draft does from what you want it to do (e.g. modelling from the
> > > >'food' example...)
> > > >The problem is in having full-text outcomes, without contextualising
> > > >structures.
> > > >So:
> > > >'Mayonnaise' will fit on a sidebar.
> > > >'Participate in collaborative conversations with diverse partners
> > > >about grade 1 topics and texts with peers and adults in small and
> > > >larger groups' will not fit on a sidebar.
> > > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > > >''Participate in collaborative conversations with diverse partners
> > > >about grade 1 topics and texts with peers and adults in small and
> > > >larger groups' exists in the semantic construct 'Outcomes'...
> > > >...but that's not actually useful information without the
> > > >contextualising specifier 'English'. (The outcome could just as easily
> > > >be taken from grade 1 'Science'.)
> > > >I note that in the string 'Participate in collaborative conversations
> > > >with diverse partners about grade 1 topics and texts with peers and
> > > >adults in small and larger groups', the string 'English' does not
> > > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > > >meaning-making fields that define the context of the outcome can be
> > > >recorded - or where the identifier for the outcome string (which ties
> > > >back to the meaning-making fields) can be stored. There's only the
> > > >full-text outcome, and as our research has shown, full-text outcomes
> > > >almost never contain strings that match user search terms.
> > > >My point here is, full-text outcome strings without context don't
> > > >support usable display of information pertaining to search results
> > > >either.
> > > >So while I think your model of 'recipe' is a good one to aim for, I
> > > >don't think that the 0.7 specification supports it just yet. That
> > > >would require identifiers and subject areas as a minimum, not an
> > > >optional extra.
> > > >Hope that made sense...
> > > >Zoe
> > > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU
> > <mailto:pdpi...@MIT.EDU>> wrote:
> > > >> Following up on Stuart's response, the model I've had in my head for
> > > >> justifying LRMI (and schema.org <http://schema.org> in general)
> > is the faceted search that
> > > >> Google now supports around recipes.
> > > >> I'm never sure if this works for everyone, but when I do a search now
> > > >>for
> > > >> "potato salad recipe" in Google, I now get a set of facets on the
> > left
> > > >> that let me narrow down my search, for ingredients, cook time and
> > > >>calorie
> > > >> count.
> > > >> As a casual recipe searcher, I think I'd be unlikely to use a
> > > >> search term "potato salad recipe with less than 300 calories"
> > much less
> > > >> "potato salad recipe with 200 calories." But having that metadata
> > > >>exposed
> > > >> to Google allows it to provide a better search experience.
> > > >> You can imagine how this might work for LRMI metadata. The user
> > starts
> > > >> with a search on "Math," but can
-- Heriot-Watt University is the Sunday Times
Scottish University of the Year 2011-2012
We invite research leaders and ambitious early career
researchers to join us in leading and driving research
in key inter-disciplinary themes. Please see
Phil, agreed. Even saying "5th grade" in the U.S. has a lot squish around
it. Any American will follow on with the question "5th grade where in the
United States"--in Beverly Hills?... somewhere where folks are less
"fortunate"? I _think_ Zoe's point comes down to the fact that most
learning resource descriptions tend to be framed in terms of some system of
education level. I believe that that's because when you ask a teacher what
she teaches, she'll respond not with a statement of ages, but with an
education level. That's also how they think and search. Ideally, the
fact that Scottish S5 maps to typical age range 14-15 is something that
should be happening more globally in the background and not repetitively
at the point of markup. We can imagine an RDFy kind of world where such
mappings exist for machines presenting me descriptions and I am not driven
to a Wikipedia page to figure out the value of a learning resource mapped
to Scottish S5 to a 9th grade teacher in the U.S. If the glue in the
mapping is typical age range--great. But Zoe's point was (I think) that a
teacher in neither Scottland nor the U.S. is likely to search for age
14-15. Of course this turns back on me in terms of my earlier post
regarding the value of the metadata outside the initial search--oh my.
On Tue, May 1, 2012 at 9:32 AM, Phil Barker <phil.bar...@hw.ac.uk> wrote:
> Age doesn't work at all well for HE either, unless you just say 18+ and
> don't differentiate between 1st year bachelor level and Master degree level.
> I think there was also an issue that grade levels are very difficult to
> map across different education systems.
> Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> Grade "usually the first year of high school" equates to Grade 8 for some
> parts of Australia and Grade 7 in others, if you map via it being "Freshman
> year", or Year 10/Form 4 in New Zealand (mapped by age), or Year 10 in
> England and Wales but Year 11 in Northern Ireland, which for some is the
> final year of High School not the first...
> At which point you perhaps think that maybe the only way to express this
> equivalence, albeit inadequately, is in terms of the typical age of the
> students? E.g.
> This resource is suitable for Scottish <span itemprop="typicalAge"
> value="14-15">S5</span> pupils
> Yes you are correct, we discussed mapping typical age to ed level. We also
> said this would not work for higher education and then sort of left it
> there. However, since we have the box of worm cans open, we might as well
> look at this educaitonalLevel issues again (Which seems like yet another
> framework alignment issue… e.g. What is your ed level framework and which
> level is this resource aligned with.) ****
> ** **
> Joshua Marks****
> CTO****
> Curriki: The Global Education and Learning Community ****
> *From:* lrmi@googlegroups.com [mailto:lrmi@googlegroups.com<lrmi@googlegroups.com>]
> *On Behalf Of *Stuart Sutton
> *Sent:* Tuesday, May 01, 2012 8:54 AM
> *To:* lrmi@googlegroups.com
> *Subject:* Re: Some thoughts from our research on curriculum models in
> metadata****
> ** **
> Zoe, dcterms:educationLevel [1] used by many Dublin Core-based projects
> for expressing the notion behind grades (in U.S. vernacular) and
> qualifications was part of the original listing of potential LRMI
> properties but was not included. I have no idea whether it is possible at
> this point to reopen those discussions/decisions since finalization of the
> LRMI specification is near.
> However, you are correct, the prevalent expressions (particularly among
> teachers/searchers) are in terms of grades/qualifications and not typical
> age range. I guess this assumes many projects will be mapping from
> education levels to typical age [see 2] or using some other non-LRMI
> property.
> On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com> wrote:****
> So, Peter, what do you recommend? Leaving things as they are, or
> modifying them to reflect the discussion?
> (On the point of pre-existing html - that's what I thought the
> intention was, but I wasn't sure as 'grade' isn't currently a field,
> only age range is. Yet very little of our learning content (and the
> learning content I see around the web) is marked up with ages, it's
> marked up with grades or exam levels. So we wouldn't be able to use
> the 'age' field.
> Additional to this - outcomes/competencies are not objectively
> appropriate for any given age, they are only meaningful in the context
> of the whole curriculum. Assuming that Texas' outcomes for eight year
> olds are the same as France's outcomes for eight year olds is asking
> for trouble - another reason why I'm starting to suspect the outcome
> context (grade, subject area, qualification) are required rather than
> optional for LRMI metadata to be meaningful.
> Zoe****
> On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > Absolutely makes sense to me.
> > In the case of relations to competencies, I don't think the text of the
> > competency is nearly as useful as the metadata around it -- first that a
> > relation exists at all; then the jurisdiction of the competency; and
> then,
> > maybe someday, the position of a competency within a curriculum.
> > Another thing to keep in mind is one of the basic ideas of microdata --
> > that the metadata shouldn't take precedence over the data. The primary
> use
> > case (at least for schema.org) is of adding semantic markup to an
> already
> > existing web page. The presumption is the text that makes the page useful
> > for humans is already there. The purpose of the microdata is to make it
> > easier for computers to read.
> > So you actually have the option of putting something sensible in the text
> > of the page, perhaps "this resource supports collaboration in grades 1-2"
> > and then back-it up with microdata that links to the competency.
> > Reply-To: "lrmi@googlegroups.com" <lrmi@googlegroups.com>
> > Date: Tuesday, May 1, 2012 8:03 AM
> > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com>
> > Subject: Re: Some thoughts from our research on curriculum models in
> > metadata
> > >Peter -
> > >I just worked out where I think the big difference if for what the
> > >LRMI draft does from what you want it to do (e.g. modelling from the
> > >'food' example...)
> > >The problem is in having full-text outcomes, without contextualising
> > >structures.
> > >So:
> > >'Mayonnaise' will fit on a sidebar.
> > >'Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' will not fit on a sidebar.
> > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > >''Participate in collaborative conversations with diverse partners
> > >about grade 1 topics and texts with peers and adults in small and
> > >larger groups' exists in the semantic construct 'Outcomes'...
> > >...but that's not actually useful information without the
> > >contextualising specifier 'English'. (The outcome could just as easily
> > >be taken from grade 1 'Science'.)
> > >I note that in the string 'Participate in collaborative conversations
> > >with diverse partners about grade 1 topics and texts with peers and
> > >adults in small and larger groups', the string 'English' does not
> > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > >meaning-making fields that define the context of the outcome can be
> > >recorded - or where the identifier for the outcome string (which ties
> > >back to the meaning-making fields) can be stored. There's only the
> > >full-text outcome, and as our research has shown, full-text outcomes
> > >almost never contain strings that match user search terms.
> > >My point here is, full-text outcome strings without context don't
> > >support usable display of information pertaining to search results
> > >either.
> > >So while I think your model of 'recipe' is a good one to aim for, I
> > >don't think that the 0.7 specification supports it just yet. That
> > >would require identifiers and subject areas as a minimum, not an
> > >optional extra.
> > >Hope that made sense...
> > >Zoe
> > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU> wrote:
> > >> Following up on Stuart's response, the model I've had in my head for
> > >> justifying LRMI (and schema.org in general) is the
Phil - correct that the grades do not map easily...
But the competencies don't map at all.
I think it's best to ignore 'map' as a concept entirely. As we've
already established, instruction has neither universal content nor
outcomes.
'Mapping' by any universal constant like 'age' gives a false sense of
uniformity at best - a class of twelve year olds in Texas will not
study the same thing as a class of twelve year olds in France, and
referencing their age as if they did will increase confusion >
decrease results relevance.
If we marked up our maths>'shape space and measure' content with '12'
and US teachers found it, it wouldn't help, it would hinder - because
we have different measurement systems.
The curricula of the world aren't standard, no, but that's a strength
not a weakness!
For metadata, it's actually a significant strength - anyone who
searches on/benefits from results markup mentioning 'KS2' has self-
identified as a UK teacher using UK outcomes. That's excellent.
I also note that the 'guiding principles' talk about using information
that's already stored/is often stored in the existing html - in all
the content I've seen/worked with, grade is a far more comment piece
of data than age.
On May 1, 5:32 pm, Phil Barker <phil.bar...@hw.ac.uk> wrote:
> Age doesn't work at all well for HE either, unless you just say 18+ and
> don't differentiate between 1st year bachelor level and Master degree level.
> I think there was also an issue that grade levels are very difficult to
> map across different education systems.
> Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> Grade "usually the first year of high school" equates to Grade 8 for
> some parts of Australia and Grade 7 in others, if you map via it being
> "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
> Year 10 in England and Wales but Year 11 in Northern Ireland, which for
> some is the final year of High School not the first...
> At which point you perhaps think that maybe the only way to express this
> equivalence, albeit inadequately, is in terms of the typical age of the
> students? E.g.
> This resource is suitable for Scottish <span itemprop="typicalAge"
> value="14-15">S5</span> pupils
> > Yes you are correct, we discussed mapping typical age to ed level. We
> > also said this would not work for higher education and then sort of
> > left it there. However, since we have the box of worm cans open, we
> > might as well look at this educaitonalLevel issues again (Which seems
> > like yet another framework alignment issue… e.g. What is your ed level
> > framework and which level is this resource aligned with.)
> > Joshua Marks
> > CTO
> > Curriki: The Global Education and Learning Community
> > *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> > Of *Stuart Sutton
> > *Sent:* Tuesday, May 01, 2012 8:54 AM
> > *To:* lrmi@googlegroups.com
> > *Subject:* Re: Some thoughts from our research on curriculum models in
> > metadata
> > Zoe, dcterms:educationLevel [1] used by many Dublin Core-based
> > projects for expressing the notion behind grades (in U.S. vernacular)
> > and qualifications was part of the original listing of potential LRMI
> > properties but was not included. I have no idea whether it is
> > possible at this point to reopen those discussions/decisions since
> > finalization of the LRMI specification is near.
> > However, you are correct, the prevalent expressions (particularly
> > among teachers/searchers) are in terms of grades/qualifications and
> > not typical age range. I guess this assumes many projects will be
> > mapping from education levels to typical age [see 2] or using some
> > other non-LRMI property.
> > On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com
> > <mailto:zoe.f.r...@gmail.com>> wrote:
> > So, Peter, what do you recommend? Leaving things as they are, or
> > modifying them to reflect the discussion?
> > (On the point of pre-existing html - that's what I thought the
> > intention was, but I wasn't sure as 'grade' isn't currently a field,
> > only age range is. Yet very little of our learning content (and the
> > learning content I see around the web) is marked up with ages, it's
> > marked up with grades or exam levels. So we wouldn't be able to use
> > the 'age' field.
> > Additional to this - outcomes/competencies are not objectively
> > appropriate for any given age, they are only meaningful in the context
> > of the whole curriculum. Assuming that Texas' outcomes for eight year
> > olds are the same as France's outcomes for eight year olds is asking
> > for trouble - another reason why I'm starting to suspect the outcome
> > context (grade, subject area, qualification) are required rather than
> > optional for LRMI metadata to be meaningful.
> > Zoe
> > On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU
> > <mailto:pdpi...@MIT.EDU>> wrote:
> > > Absolutely makes sense to me.
> > > In the case of relations to competencies, I don't think the text of the
> > > competency is nearly as useful as the metadata around it -- first that a
> > > relation exists at all; then the jurisdiction of the competency; and
> > then,
> > > maybe someday, the position of a competency within a curriculum.
> > > Another thing to keep in mind is one of the basic ideas of microdata --
> > > that the metadata shouldn't take precedence over the data. The
> > primary use
> > > case (at least for schema.org <http://schema.org>) is of adding
> > semantic markup to an already
> > > existing web page. The presumption is the text that makes the page
> > useful
> > > for humans is already there. The purpose of the microdata is to make it
> > > easier for computers to read.
> > > So you actually have the option of putting something sensible in the
> > text
> > > of the page, perhaps "this resource supports collaboration in grades
> > 1-2"
> > > and then back-it up with microdata that links to the competency.
> > > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>"
> > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > > Date: Tuesday, May 1, 2012 8:03 AM
> > > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com
> > <mailto:lrmi@googlegroups.com>>
> > > Subject: Re: Some thoughts from our research on curriculum models in
> > > metadata
> > > >Peter -
> > > >I just worked out where I think the big difference if for what the
> > > >LRMI draft does from what you want it to do (e.g. modelling from the
> > > >'food' example...)
> > > >The problem is in having full-text outcomes, without contextualising
> > > >structures.
> > > >So:
> > > >'Mayonnaise' will fit on a sidebar.
> > > >'Participate in collaborative conversations with diverse partners
> > > >about grade 1 topics and texts with peers and adults in small and
> > > >larger groups' will not fit on a sidebar.
> > > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > > >''Participate in collaborative conversations with diverse partners
> > > >about grade 1 topics and texts with peers and adults in small and
> > > >larger groups' exists in the semantic construct 'Outcomes'...
> > > >...but that's not actually useful information without the
> > > >contextualising specifier 'English'. (The outcome could just as easily
> > > >be taken from grade 1 'Science'.)
> > > >I note that in the string 'Participate in collaborative conversations
> > > >with diverse partners about grade 1 topics and texts with peers and
> > > >adults in small and larger groups', the string 'English' does not
> > > >occur - nor does 'speaking', 'listening', or 'speaking and listening'.
> > > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > > >meaning-making fields that define the context of the outcome can be
> > > >recorded - or where the identifier for the outcome string (which ties
> > > >back to the meaning-making fields) can be stored. There's only the
> > > >full-text outcome, and as our research has shown, full-text outcomes
> > > >almost never contain strings that match user search terms.
> > > >My point here is, full-text outcome strings without context don't
> > > >support usable display of information pertaining to search results
> > > >either.
> > > >So while I think your model of 'recipe' is a good one to aim for, I
> > > >don't think that the 0.7 specification supports it just yet. That
> > > >would require identifiers and subject areas as a minimum, not an
> > > >optional extra.
> On May 1, 5:32 pm, Phil Barker <phil.bar...@hw.ac.uk> wrote:
> > Age doesn't work at all well for HE either, unless you just say 18+ and
> > don't differentiate between 1st year bachelor level and Master degree
> level.
> > I think there was also an issue that grade levels are very difficult to
> > map across different education systems.
> > Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> > Grade "usually the first year of high school" equates to Grade 8 for
> > some parts of Australia and Grade 7 in others, if you map via it being
> > "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
> > Year 10 in England and Wales but Year 11 in Northern Ireland, which for
> > some is the final year of High School not the first...
> > At which point you perhaps think that maybe the only way to express this
> > equivalence, albeit inadequately, is in terms of the typical age of the
> > students? E.g.
> > This resource is suitable for Scottish <span itemprop="typicalAge"
> > value="14-15">S5</span> pupils
> > > Yes you are correct, we discussed mapping typical age to ed level. We
> > > also said this would not work for higher education and then sort of
> > > left it there. However, since we have the box of worm cans open, we
> > > might as well look at this educaitonalLevel issues again (Which seems
> > > like yet another framework alignment issue… e.g. What is your ed level
> > > framework and which level is this resource aligned with.)
> > > Joshua Marks
> > > CTO
> > > Curriki: The Global Education and Learning Community
> > > *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> > > Of *Stuart Sutton
> > > *Sent:* Tuesday, May 01, 2012 8:54 AM
> > > *To:* lrmi@googlegroups.com
> > > *Subject:* Re: Some thoughts from our research on curriculum models in
> > > metadata
> > > Zoe, dcterms:educationLevel [1] used by many Dublin Core-based
> > > projects for expressing the notion behind grades (in U.S. vernacular)
> > > and qualifications was part of the original listing of potential LRMI
> > > properties but was not included. I have no idea whether it is
> > > possible at this point to reopen those discussions/decisions since
> > > finalization of the LRMI specification is near.
> > > However, you are correct, the prevalent expressions (particularly
> > > among teachers/searchers) are in terms of grades/qualifications and
> > > not typical age range. I guess this assumes many projects will be
> > > mapping from education levels to typical age [see 2] or using some
> > > other non-LRMI property.
> > > On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com
> > > <mailto:zoe.f.r...@gmail.com>> wrote:
> > > So, Peter, what do you recommend? Leaving things as they are, or
> > > modifying them to reflect the discussion?
> > > (On the point of pre-existing html - that's what I thought the
> > > intention was, but I wasn't sure as 'grade' isn't currently a field,
> > > only age range is. Yet very little of our learning content (and the
> > > learning content I see around the web) is marked up with ages, it's
> > > marked up with grades or exam levels. So we wouldn't be able to use
> > > the 'age' field.
> > > Additional to this - outcomes/competencies are not objectively
> > > appropriate for any given age, they are only meaningful in the context
> > > of the whole curriculum. Assuming that Texas' outcomes for eight year
> > > olds are the same as France's outcomes for eight year olds is asking
> > > for trouble - another reason why I'm starting to suspect the outcome
> > > context (grade, subject area, qualification) are required rather than
> > > optional for LRMI metadata to be meaningful.
> > > Zoe
> > > On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU
> > > <mailto:pdpi...@MIT.EDU>> wrote:
> > > > Absolutely makes sense to me.
> > > > In the case of relations to competencies, I don't think the text of
> the
> > > > competency is nearly as useful as the metadata around it -- first
> that a
> > > > relation exists at all; then the jurisdiction of the competency; and
> > > then,
> > > > maybe someday, the position of a competency within a curriculum.
> > > > Another thing to keep in mind is one of the basic ideas of microdata
> --
> > > > that the metadata shouldn't take precedence over the data. The
> > > primary use
> > > > case (at least for schema.org <http://schema.org>) is of adding
> > > semantic markup to an already
> > > > existing web page. The presumption is the text that makes the page
> > > useful
> > > > for humans is already there. The purpose of the microdata is to make
> it
> > > > easier for computers to read.
> > > > So you actually have the option of putting something sensible in the
> > > text
> > > > of the page, perhaps "this resource supports collaboration in grades
> > > 1-2"
> > > > and then back-it up with microdata that links to the competency.
> > > > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>"
> > > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > > > Date: Tuesday, May 1, 2012 8:03 AM
> > > > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com
> > > <mailto:lrmi@googlegroups.com>>
> > > > Subject: Re: Some thoughts from our research on curriculum models in
> > > > metadata
> > > > >Peter -
> > > > >I just worked out where I think the big difference if for what the
> > > > >LRMI draft does from what you want it to do (e.g. modelling from the
> > > > >'food' example...)
> > > > >The problem is in having full-text outcomes, without contextualising
> > > > >structures.
> > > > >So:
> > > > >'Mayonnaise' will fit on a sidebar.
> > > > >'Participate in collaborative conversations with diverse partners
> > > > >about grade 1 topics and texts with peers and adults in small and
> > > > >larger groups' will not fit on a sidebar.
> > > > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > > > >''Participate in collaborative conversations with diverse partners
> > > > >about grade 1 topics and texts with peers and adults in small and
> > > > >larger groups' exists in the semantic construct 'Outcomes'...
> > > > >...but that's not actually useful information without the
> > > > >contextualising specifier 'English'. (The outcome could just as
> easily
> > > > >be taken from grade 1 'Science'.)
> > > > >I note that in the string 'Participate in collaborative
> conversations
> > > > >with diverse partners about grade 1 topics and texts with peers and
> > > > >adults in small and larger groups', the string 'English' does not
> > > > >occur - nor does 'speaking', 'listening', or 'speaking and
> listening'.
> > > > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > > > >meaning-making fields that define the context of the outcome can be
> > > > >recorded - or where the identifier for the outcome string (which
> ties
> > > > >back to the meaning-making fields) can be stored. There's only the
> > > > >full-text outcome, and as our research has shown, full-text outcomes
> > > > >almost never contain strings that match user search terms.
> > > > >My point here is, full-text outcome strings without context don't
> > > > >support usable display of information pertaining to search results
> > > > >either.
> > > > >So while I think your model of 'recipe' is a good one to aim for, I
> > > > >don't think that the 0.7 specification supports it just yet. That
> > > > >would require identifiers and subject areas as a minimum, not an
> > > > >optional extra.
> > > > >Hope that made sense...
> > > > >Zoe
> > > > >On Apr 30, 3:10 pm, Peter Pinch <pdpi...@MIT.EDU
> > > <mailto:pdpi...@MIT.EDU>> wrote:
> > > > >> Following up on Stuart's response, the model I've had in my head
> for
> > > > >> justifying LRMI (and schema.org <http://schema.org> in general)
> > > is the faceted search that
> > > > >> Google now supports around recipes.
> > > > >> I'm never sure if this works for everyone, but when I do a search
> now
> > > > >>for
> > > > >> "potato salad
My point is that ages are universal for students (they all turn twelve
at some point), but not for what they get taught.
It's sticky, but the complexity is unavoidable - students learn
different things at different ages depending on what school system
they're taught in.
I also point back to the outcomes - those are organised by grade, so,
they're only meaningful when associated with a grade (and subject
area). As full text, decontextualised strings, there's not much you
can do with them.
On May 1, 6:06 pm, Stuart Sutton <stuartasut...@gmail.com> wrote:
> On Tue, May 1, 2012 at 9:53 AM, Zoe <zoe.f.r...@gmail.com> wrote:
> > Wait - Phil, is the scope of the LRMI supposed to encompass HE and FE?
> > As far as I can tell the LRMI scope is for learning resource markup for
> search/discover. No limitations on the context.
> Stuart
> > On May 1, 5:32 pm, Phil Barker <phil.bar...@hw.ac.uk> wrote:
> > > Age doesn't work at all well for HE either, unless you just say 18+ and
> > > don't differentiate between 1st year bachelor level and Master degree
> > level.
> > > I think there was also an issue that grade levels are very difficult to
> > > map across different education systems.
> > > Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> > > Grade "usually the first year of high school" equates to Grade 8 for
> > > some parts of Australia and Grade 7 in others, if you map via it being
> > > "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
> > > Year 10 in England and Wales but Year 11 in Northern Ireland, which for
> > > some is the final year of High School not the first...
> > > At which point you perhaps think that maybe the only way to express this
> > > equivalence, albeit inadequately, is in terms of the typical age of the
> > > students? E.g.
> > > This resource is suitable for Scottish <span itemprop="typicalAge"
> > > value="14-15">S5</span> pupils
> > > > Yes you are correct, we discussed mapping typical age to ed level. We
> > > > also said this would not work for higher education and then sort of
> > > > left it there. However, since we have the box of worm cans open, we
> > > > might as well look at this educaitonalLevel issues again (Which seems
> > > > like yet another framework alignment issue… e.g. What is your ed level
> > > > framework and which level is this resource aligned with.)
> > > > Joshua Marks
> > > > CTO
> > > > Curriki: The Global Education and Learning Community
> > > > *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> > > > Of *Stuart Sutton
> > > > *Sent:* Tuesday, May 01, 2012 8:54 AM
> > > > *To:* lrmi@googlegroups.com
> > > > *Subject:* Re: Some thoughts from our research on curriculum models in
> > > > metadata
> > > > Zoe, dcterms:educationLevel [1] used by many Dublin Core-based
> > > > projects for expressing the notion behind grades (in U.S. vernacular)
> > > > and qualifications was part of the original listing of potential LRMI
> > > > properties but was not included. I have no idea whether it is
> > > > possible at this point to reopen those discussions/decisions since
> > > > finalization of the LRMI specification is near.
> > > > However, you are correct, the prevalent expressions (particularly
> > > > among teachers/searchers) are in terms of grades/qualifications and
> > > > not typical age range. I guess this assumes many projects will be
> > > > mapping from education levels to typical age [see 2] or using some
> > > > other non-LRMI property.
> > > > On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com
> > > > <mailto:zoe.f.r...@gmail.com>> wrote:
> > > > So, Peter, what do you recommend? Leaving things as they are, or
> > > > modifying them to reflect the discussion?
> > > > (On the point of pre-existing html - that's what I thought the
> > > > intention was, but I wasn't sure as 'grade' isn't currently a field,
> > > > only age range is. Yet very little of our learning content (and the
> > > > learning content I see around the web) is marked up with ages, it's
> > > > marked up with grades or exam levels. So we wouldn't be able to use
> > > > the 'age' field.
> > > > Additional to this - outcomes/competencies are not objectively
> > > > appropriate for any given age, they are only meaningful in the context
> > > > of the whole curriculum. Assuming that Texas' outcomes for eight year
> > > > olds are the same as France's outcomes for eight year olds is asking
> > > > for trouble - another reason why I'm starting to suspect the outcome
> > > > context (grade, subject area, qualification) are required rather than
> > > > optional for LRMI metadata to be meaningful.
> > > > Zoe
> > > > On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU
> > > > <mailto:pdpi...@MIT.EDU>> wrote:
> > > > > Absolutely makes sense to me.
> > > > > In the case of relations to competencies, I don't think the text of
> > the
> > > > > competency is nearly as useful as the metadata around it -- first
> > that a
> > > > > relation exists at all; then the jurisdiction of the competency; and
> > > > then,
> > > > > maybe someday, the position of a competency within a curriculum.
> > > > > Another thing to keep in mind is one of the basic ideas of microdata
> > --
> > > > > that the metadata shouldn't take precedence over the data. The
> > > > primary use
> > > > > case (at least for schema.org <http://schema.org>) is of adding
> > > > semantic markup to an already
> > > > > existing web page. The presumption is the text that makes the page
> > > > useful
> > > > > for humans is already there. The purpose of the microdata is to make
> > it
> > > > > easier for computers to read.
> > > > > So you actually have the option of putting something sensible in the
> > > > text
> > > > > of the page, perhaps "this resource supports collaboration in grades
> > > > 1-2"
> > > > > and then back-it up with microdata that links to the competency.
> > > > > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>"
> > > > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > > > > Date: Tuesday, May 1, 2012 8:03 AM
> > > > > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com
> > > > <mailto:lrmi@googlegroups.com>>
> > > > > Subject: Re: Some thoughts from our research on curriculum models in
> > > > > metadata
> > > > > >Peter -
> > > > > >I just worked out where I think the big difference if for what the
> > > > > >LRMI draft does from what you want it to do (e.g. modelling from the
> > > > > >'food' example...)
> > > > > >The problem is in having full-text outcomes, without contextualising
> > > > > >structures.
> > > > > >So:
> > > > > >'Mayonnaise' will fit on a sidebar.
> > > > > >'Participate in collaborative conversations with diverse partners
> > > > > >about grade 1 topics and texts with peers and adults in small and
> > > > > >larger groups' will not fit on a sidebar.
> > > > > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > > > > >''Participate in collaborative conversations with diverse partners
> > > > > >about grade 1 topics and texts with peers and adults in small and
> > > > > >larger groups' exists in the semantic construct 'Outcomes'...
> > > > > >...but that's not actually useful information without the
> > > > > >contextualising specifier 'English'. (The outcome could just as
> > easily
> > > > > >be taken from grade 1 'Science'.)
> > > > > >I note that in the string 'Participate in collaborative
> > conversations
> > > > > >with diverse partners about grade 1 topics and texts with peers and
> > > > > >adults in small and larger groups', the string 'English' does not
> > > > > >occur - nor does 'speaking', 'listening', or 'speaking and
> > listening'.
> > > > > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > > > > >meaning-making fields that define the context of the outcome can be
> > > > > >recorded - or where the identifier for the outcome string (which
> > ties
> > > > > >back to the meaning-making fields) can be stored. There's only the
> > > > > >full-text outcome, and as our research has shown, full-text outcomes
> > > > > >almost never contain strings that match user search terms.
> > > > > >My point here is, full-text outcome strings without context don't
> > > > > >support usable display of information pertaining to search results
HE education is an area of interest of mine (it's what you get for
marrying an academic ;) )
As a first impression, I'm not sure that the current LRMI fields would
support the breadth of teaching currently undertaken in FE and HE.
If FE and HE are on the list, then 'Qualification' becomes really
important (especially for FE as Qualification is how all their
learning is organised) and 'Subject area' even more so.
Zoe
On May 1, 6:06 pm, Stuart Sutton <stuartasut...@gmail.com> wrote:
> On Tue, May 1, 2012 at 9:53 AM, Zoe <zoe.f.r...@gmail.com> wrote:
> > Wait - Phil, is the scope of the LRMI supposed to encompass HE and FE?
> > As far as I can tell the LRMI scope is for learning resource markup for
> search/discover. No limitations on the context.
> Stuart
> > On May 1, 5:32 pm, Phil Barker <phil.bar...@hw.ac.uk> wrote:
> > > Age doesn't work at all well for HE either, unless you just say 18+ and
> > > don't differentiate between 1st year bachelor level and Master degree
> > level.
> > > I think there was also an issue that grade levels are very difficult to
> > > map across different education systems.
> > > Have a look the wikipedia entry on Ninth Grade [1] and see how US 9th
> > > Grade "usually the first year of high school" equates to Grade 8 for
> > > some parts of Australia and Grade 7 in others, if you map via it being
> > > "Freshman year", or Year 10/Form 4 in New Zealand (mapped by age), or
> > > Year 10 in England and Wales but Year 11 in Northern Ireland, which for
> > > some is the final year of High School not the first...
> > > At which point you perhaps think that maybe the only way to express this
> > > equivalence, albeit inadequately, is in terms of the typical age of the
> > > students? E.g.
> > > This resource is suitable for Scottish <span itemprop="typicalAge"
> > > value="14-15">S5</span> pupils
> > > > Yes you are correct, we discussed mapping typical age to ed level. We
> > > > also said this would not work for higher education and then sort of
> > > > left it there. However, since we have the box of worm cans open, we
> > > > might as well look at this educaitonalLevel issues again (Which seems
> > > > like yet another framework alignment issue… e.g. What is your ed level
> > > > framework and which level is this resource aligned with.)
> > > > Joshua Marks
> > > > CTO
> > > > Curriki: The Global Education and Learning Community
> > > > *From:*lrmi@googlegroups.com [mailto:lrmi@googlegroups.com] *On Behalf
> > > > Of *Stuart Sutton
> > > > *Sent:* Tuesday, May 01, 2012 8:54 AM
> > > > *To:* lrmi@googlegroups.com
> > > > *Subject:* Re: Some thoughts from our research on curriculum models in
> > > > metadata
> > > > Zoe, dcterms:educationLevel [1] used by many Dublin Core-based
> > > > projects for expressing the notion behind grades (in U.S. vernacular)
> > > > and qualifications was part of the original listing of potential LRMI
> > > > properties but was not included. I have no idea whether it is
> > > > possible at this point to reopen those discussions/decisions since
> > > > finalization of the LRMI specification is near.
> > > > However, you are correct, the prevalent expressions (particularly
> > > > among teachers/searchers) are in terms of grades/qualifications and
> > > > not typical age range. I guess this assumes many projects will be
> > > > mapping from education levels to typical age [see 2] or using some
> > > > other non-LRMI property.
> > > > On Tue, May 1, 2012 at 8:33 AM, Zoe <zoe.f.r...@gmail.com
> > > > <mailto:zoe.f.r...@gmail.com>> wrote:
> > > > So, Peter, what do you recommend? Leaving things as they are, or
> > > > modifying them to reflect the discussion?
> > > > (On the point of pre-existing html - that's what I thought the
> > > > intention was, but I wasn't sure as 'grade' isn't currently a field,
> > > > only age range is. Yet very little of our learning content (and the
> > > > learning content I see around the web) is marked up with ages, it's
> > > > marked up with grades or exam levels. So we wouldn't be able to use
> > > > the 'age' field.
> > > > Additional to this - outcomes/competencies are not objectively
> > > > appropriate for any given age, they are only meaningful in the context
> > > > of the whole curriculum. Assuming that Texas' outcomes for eight year
> > > > olds are the same as France's outcomes for eight year olds is asking
> > > > for trouble - another reason why I'm starting to suspect the outcome
> > > > context (grade, subject area, qualification) are required rather than
> > > > optional for LRMI metadata to be meaningful.
> > > > Zoe
> > > > On May 1, 2:07 pm, Peter Pinch <pdpi...@MIT.EDU
> > > > <mailto:pdpi...@MIT.EDU>> wrote:
> > > > > Absolutely makes sense to me.
> > > > > In the case of relations to competencies, I don't think the text of
> > the
> > > > > competency is nearly as useful as the metadata around it -- first
> > that a
> > > > > relation exists at all; then the jurisdiction of the competency; and
> > > > then,
> > > > > maybe someday, the position of a competency within a curriculum.
> > > > > Another thing to keep in mind is one of the basic ideas of microdata
> > --
> > > > > that the metadata shouldn't take precedence over the data. The
> > > > primary use
> > > > > case (at least for schema.org <http://schema.org>) is of adding
> > > > semantic markup to an already
> > > > > existing web page. The presumption is the text that makes the page
> > > > useful
> > > > > for humans is already there. The purpose of the microdata is to make
> > it
> > > > > easier for computers to read.
> > > > > So you actually have the option of putting something sensible in the
> > > > text
> > > > > of the page, perhaps "this resource supports collaboration in grades
> > > > 1-2"
> > > > > and then back-it up with microdata that links to the competency.
> > > > > Reply-To: "lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>"
> > > > <lrmi@googlegroups.com <mailto:lrmi@googlegroups.com>>
> > > > > Date: Tuesday, May 1, 2012 8:03 AM
> > > > > To: Learning Resource Metadata Initiative <lrmi@googlegroups.com
> > > > <mailto:lrmi@googlegroups.com>>
> > > > > Subject: Re: Some thoughts from our research on curriculum models in
> > > > > metadata
> > > > > >Peter -
> > > > > >I just worked out where I think the big difference if for what the
> > > > > >LRMI draft does from what you want it to do (e.g. modelling from the
> > > > > >'food' example...)
> > > > > >The problem is in having full-text outcomes, without contextualising
> > > > > >structures.
> > > > > >So:
> > > > > >'Mayonnaise' will fit on a sidebar.
> > > > > >'Participate in collaborative conversations with diverse partners
> > > > > >about grade 1 topics and texts with peers and adults in small and
> > > > > >larger groups' will not fit on a sidebar.
> > > > > >'Mayonnnaise' exists in the semantic construct 'ingredients'
> > > > > >''Participate in collaborative conversations with diverse partners
> > > > > >about grade 1 topics and texts with peers and adults in small and
> > > > > >larger groups' exists in the semantic construct 'Outcomes'...
> > > > > >...but that's not actually useful information without the
> > > > > >contextualising specifier 'English'. (The outcome could just as
> > easily
> > > > > >be taken from grade 1 'Science'.)
> > > > > >I note that in the string 'Participate in collaborative
> > conversations
> > > > > >with diverse partners about grade 1 topics and texts with peers and
> > > > > >adults in small and larger groups', the string 'English' does not
> > > > > >occur - nor does 'speaking', 'listening', or 'speaking and
> > listening'.
> > > > > >I also note that the LRMI 0.7 draft doesn't have an area where these
> > > > > >meaning-making fields that define the context of the outcome can be
> > > > > >recorded - or where the identifier for the outcome string (which
> > ties
> > > > > >back to the meaning-making fields) can be stored. There's only the
> > > > > >full-text outcome, and as our research has shown, full-text outcomes
> > > > > >almost never contain strings that match user search terms.
> > > > > >My point here is, full-text outcome strings without context don't
> > > > > >support usable display of information pertaining to search results
> > > > > >either.
> > > > > >So while I think your model of 'recipe' is a good one to aim
Probably we all recognise that
(a) what is taught at a particular age varies not only between different
countries, but also in different educational subsystems within a country,
and in the same country in different years;
(b) there is little correlation between school "years", whether or not they
have the same number (in any case compulsory education starts at different
ages).
What is relatively well defined is what is expected within a certain
(perhaps national or state) curriculum. That's what is published, what
authors and book companies aim at, etc. To my way of thinking, there is no
harm at all in referring to a defined level within a particular educational
level system at a particular date. It is open to anyone who wants to do
linked data to propose equivalences, either across the curriculum or in
specific subject areas. When equivalences become at least mutually, or
ideally more widely, recognised, mapping across different educational
systems can begin.
Where age is actually relevant seems subtly different. Is there at least
more of a general consensus about broad levels of social/emotional
maturity, relating to how films/movies/games etc. are age rated?
Competence is more of a specialist area for me, so I'd better reply
separately on that. I hope, more in reply to different points...
Simon
On 1 May 2012 18:06, Zoe <zoe.f.r...@gmail.com> wrote:
> Phil - correct that the grades do not map easily...
> But the competencies don't map at all.
> I think it's best to ignore 'map' as a concept entirely. As we've
> already established, instruction has neither universal content nor
> outcomes.
> 'Mapping' by any universal constant like 'age' gives a false sense of
> uniformity at best - a class of twelve year olds in Texas will not
> study the same thing as a class of twelve year olds in France, and
> referencing their age as if they did will increase confusion >
> decrease results relevance.
> If we marked up our maths>'shape space and measure' content with '12'
> and US teachers found it, it wouldn't help, it would hinder - because
> we have different measurement systems.
> The curricula of the world aren't standard, no, but that's a strength
> not a weakness!
> For metadata, it's actually a significant strength - anyone who
> searches on/benefits from results markup mentioning 'KS2' has self-
> identified as a UK teacher using UK outcomes. That's excellent.
> I also note that the 'guiding principles' talk about using information
> that's already stored/is often stored in the existing html - in all
> the content I've seen/worked with, grade is a far more comment piece
> of data than age.
> Phil - correct that the grades do not map easily...
> But the competencies don't map at all.
> I think it's best to ignore 'map' as a concept entirely. As we've
> already established, instruction has neither universal content nor
> outcomes.
> 'Mapping' by any universal constant like 'age' gives a false sense of
> uniformity at best - a class of twelve year olds in Texas will not
> study the same thing as a class of twelve year olds in France, and
> referencing their age as if they did will increase confusion>
> decrease results relevance.
> If we marked up our maths>'shape space and measure' content with '12'
> and US teachers found it, it wouldn't help, it would hinder - because
> we have different measurement systems.
Hmm, let's keep age and competencies separate. I don't think anyone is suggesting that we would map competencies (not at least within LRMI--there might be some work on state -> common core mappings but that is outwith LRMI), or that we would mark up a competency statement with an age. That's why they are separate properties.
> The curricula of the world aren't standard, no, but that's a strength
> not a weakness!
> For metadata, it's actually a significant strength - anyone who
> searches on/benefits from results markup mentioning 'KS2' has self-
> identified as a UK teacher using UK outcomes. That's excellent.
> I also note that the 'guiding principles' talk about using information
> that's already stored/is often stored in the existing html - in all
> the content I've seen/worked with, grade is a far more comment piece
> of data than age.
I wouldn't be against wrapping the text KS2[1] in mark-up to say that it was an educational level but I'm not sure it adds much in the scenario that you describe.
However, I think that someone searching for KS2 is searching for materials that meet the National Curriculum requirements for Key Stage 2. Going back to Jim Goodell's suggested definition of thing to which the property currently known as competency expresses an alignment
> "Content that either describes [and references] a specific competency
> (learning objective) or describes [and references] a [more granular]
> grouping of competencies within the taxonomy of a Learning Standards
> Document [or framework]."
would that (perhaps with some tweaking) allow an expression that the material is suitable for KS2? It aligns with (e.g. teaches) the "grouping of competencies [read: objectives or something similar] within a learning [...] framework" that is known as KS2. I'm not saying that KS2 is a competency, because I don't think that 'competency' fully describes what the property covers.
