[LRMI] Zoe's proposal

[Greg Grossmeier]

Hello LRMI followers

I've taken the liberty to share with you Zoe's proposal on what changes she believes should be made to the LRMI spec to deal with the situation she has experience in (specifically, the UK educational system).

Please provide any feedback you have as soon as you can.

Thanks,

Greg

---------- Forwarded message ----------
From: Zoe Rose <zoe.f...@gmail.com>
Date: Thu, Aug 23, 2012 at 10:46 AM
Subject: Re: Fwd: I just figured it out - the problem with relying on URIs for schools-based learning content
To: Greg Grossmeier <gr...@creativecommons.org>


Hi Greg!

It is 6.12 pm on my last day at work and I refuse to turn off the computer before sending you my extremely late answer to your very simple question:

"Given your concerns, can you give me a schema.org like description of what you would like alignmentObject to contain to meet your need? Specific terms with good-enough definitions would be great."

Okie dokes:

Term: educationalAuthority

Definition: the authority which defines the other attributes in alignmentObject

Example: Royal Life Saving Society, PRINCE2, Welsh national curriculum

Because: the other descriptive statements are only meaningful in the context of the defining body. Full explanation here https://groups.google.com/group/lrmi/msg/ac6e2f80e72054b5?dmode=source 

_____________________________________________

Term: educationalLevel

Definition: the stage (not level of achievement) of learning

Example: Piano Grade seven, GCSE, NVQ, Diploma, Year three

Because: These are meaningful instances (when defined by an educationalAuthority) that are sufficiently semantically distinct from the other content in alignmentType that they can stand alone. They are not capable of conveying useful meaning when divorced from educationalAuthority -  - e.g. 'Diploma' is a term used in both High Schools and Universities in the UK, and 'Year three' refers to a wide range of different ages across the different nations. Full explanation here: https://groups.google.com/group/lrmi/msg/ded41c134be9157c?dmode=source

_____________________________________________

Term: educationalSubject

Definition: the administrative grouping of content determined by the educationalAuthority

Example: Science, the World Around Us (*that's taught in Northern Ireland, it's Science and Social Studies basically), ESOL, equine dentistry

Because: Each authority has it's own administrative groupings, which may - or may not - have crossover. (See prev . links for explanation of why curriculum mapping by naturally occuring strings doesn't work.) It is not the case that there is a 'universal curriculum' in which physics = physics = physics = 'force and motion', unfortunately.

It is also not the case that 'outcomes' have intrinsic relationships with 'content' (see: earlier in this email thread).

_____________________________________________

Why I don't think 'targetDescription' does the job:

- it is not meaningful to a non-expert (what is 'target'?)

- there is no definition for 'node' (I do not believe that any definition of 'node' can be meaningful enough to cater to all use cases discussed by the group)

_____________________________________________

Why I don't think 'targetName' does the job:

- it is not meaningful to a non-expert (what is 'target'?)

- to the extent that the concept 'node' is meaningful (see previous comment), 'nodes' do not, in most curriculum/learning documentation, have 'names'. The closest you're likely to get is outcomes, and outcomes do not describe content (see above).

_____________________________________________

Why I don't think 'targetURL' does the job

- just joking, targetURL does a fine job with URLs :)

- ...but having said that - there aren't enough stable/authoritative/cool URLs within 'established educational frameworks' for targetURL to carry all the weight just yet (and even if there were, our best content producers - teachers and small players - would not know how to use them and would be excluded) - which is, of course, why I raised this as an issue in the first place :)

(FWIW - textComplexity and readingLevel were well-intentioned but ultimately failed ideas from the LOM specification - they're kind of zombie-y. I'd suggest that LRMI would be a healthier spec without them, but it's not a showstopper.)

Yours in tardiness,

Zoe

[Greg Grossmeier]

My response to the proposal below:

> ---------- Forwarded message ----------
> From: Zoe Rose <zoe.f...@gmail.com>
>

> *Term:* educationalAuthority
>
> *Definition:* the authority which defines the other attributes in
> alignmentObject
>
> *Example:* Royal Life Saving Society, PRINCE2, Welsh national curriculum
>
> *Because:* the other descriptive statements are only meaningful in the

> context of the defining body. Full explanation here
> https://groups.google.com/group/lrmi/msg/ac6e2f80e72054b5?dmode=source

Agreed. This proposal is pretty well agreed upon (at least, no one has
voiced a concern over my iteration of it on the mailing list:
https://groups.google.com/forum/#!msg/lrmi/OjpnZ8s_He8/jTqHKk4vVUAJ )

> *Term:* educationalLevel
>
> *Definition:* the stage (not level of achievement) of learning
>
> *Example:* Piano Grade seven, GCSE, NVQ, Diploma, Year three

This is addressed by alignmentType. From the definition of
alignmentType: "A category of alignment between the learning resource
and the framework node. Recommended values include: 'assessed',
'teaches', 'requires', 'textComplexity', 'readingLevel',
'educationalSubject', and 'educationLevel'."

> *Term:* educationalSubject
>
> *Definition:* the administrative grouping of content determined by the
> educationalAuthority
>
> *Example:* Science, the World Around Us (*that's taught in Northern

> Ireland, it's Science and Social Studies basically), ESOL, equine dentistry

Ditto from above.


So, I believe we can move forward with adding the educationalAthority
term to alignmentObject.

Thanks,

Greg

--
Greg Grossmeier
Education Technology & Policy Coordinator
twitter: @g_gerg / identi.ca: @greg / skype: greg.grossmeier

[Joshua Marks]

+1 to your comments and proposal, and a wonderful "Thank you" to Zoe for her
clarity and persistence.

Joshua Marks
CTO
Curriki: The Global Education and Learning Community
jma...@curriki.org
www.curriki.org
US 831-685-3511

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.

[Dan Brickley]

Term: educationalAuthority

Definition: the authority which defines the other attributes in alignmentObject

Example: Royal Life Saving Society, PRINCE2, Welsh national curriculum

Because: the other descriptive statements are only meaningful in the context of the defining body

I'm fine with simple text labels here, but would also like to be able to use any http://schema.org/Organization ( often but not necessarily a http://schema.org/EducationalOrganization). This would allow other attributes of the org (including identifiers) to be included.

Dan (via phone)

[Peter Pinch]

+1 on Dan's suggestion

From: Dan Brickley <dan...@danbri.org>
Reply-To: "lr...@googlegroups.com" <lr...@googlegroups.com>
Date: Saturday, August 25, 2012 12:45 AM
To: "lr...@googlegroups.com" <lr...@googlegroups.com>
Subject: Re: [LRMI] Zoe's proposal

[Stuart Sutton]

I have one question of the proposers of educationalAuthority that impacts your point, Dan.  What _exactly_ is meant by "authority"?  Neither the definition nor the examples answer the question definitively (or, at least, not for me).  The definition and examples conflate: (1) what others consider the jurisdictional authority of the "framework"--e.g.,

• "Northern Ireland"
  
• "Royal Life Saving Society"
  
• "North Carolina, State Board of Education, Department of Instruction"
  
• "American Association for the Advancement of Science"
  

i.e., some governmental or other organization promulgating the framework and having authority over it (perhaps even over its application within a political region); and (2) the authoritative framework itself--using the proposers example, the "Welsh National Curriculum". 

So, one possible meaning is that educationalAuthority identifies an authoritative _organization_ or _political entity_; and, another quite distinct and useful meaning denotes identification of an authoritative _text_  (using "text" loosely to mean any form of authoritative expression of the framework as set out by its creators).  The distinction between these two meanings is sufficiently non-trivial that in most representations of such frameworks, they are distinct assertions.  Use of http://schema.org/Organization would be appropriate with one reading of the term's intent and possibly not the other (or at least I don't think so). 

Stuart

--
Stuart A. Sutton,
CEO and Managing Director, Dublin Core Metadata Initiative
Associate Professor Emeritus, The Information School
University of Washington

[Paul Libbrecht]

*Term:* educationalAuthority

*Definition:* the authority which defines the other attributes in

alignmentObject

*Example:* Royal Life Saving Society, PRINCE2, Welsh national curriculum

Agreed. This proposal is pretty well agreed upon (at least, no one has
voiced a concern over my iteration of it on the mailing list:
https://groups.google.com/forum/#!msg/lrmi/OjpnZ8s_He8/jTqHKk4vVUAJ )

Le 25 août 2012 à 06:45, Dan Brickley a écrit :

I'm fine with simple text labels here, but would also like to be able to use any http://schema.org/Organization ( often but not necessarily a http://schema.org/EducationalOrganization). This would allow other attributes of the org (including identifiers) to be included.

Fully agree this would be helpful.

About the following: Is this agreed upon that educationLevel includes an entity with URL and text, just in case there are systems that already carry URL-based identifications of levels?

paul

[Phil Barker]

Good points from Stuart. Looking back at the email that Zoe references[1], I think the ambiguity that she objects to arises from using AlignmentObject to identify a node in an educational framework with a text label without identifying the educational framework. So it is the Framework that needs to be identified, e.g. "Welsh National Curriculum" etc. Identifying the organization that publishes/promulgates/sponsors the framework won't help if that organization is responsible for >1 frameworks.

I agree with Greg's response to Zoe's other proposals.

many thanks to Zoe.


Phil


1.  https://groups.google.com/group/lrmi/msg/ac6e2f80e72054b5?dmode=source

-- 
<http://www.icbl.hw.ac.uk/~philb/>
Please note new email address: phil....@hw.ac.uk

---

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 www.hw.ac.uk/researchleaders for further information and how to apply.

Heriot-Watt University is a Scottish charity registered under charity number SC000278.

[Brandt Redd]

I like where we're going with this. I agreed with Dan's suggestion that educationalAuthority could be an organization but I was concerned that the markup would get pretty complex with alignmentObject (an abstract type) incorporating Organization (another abstract type). This is logically really clean but the syntax of nested types may be beyond the average webmaster.

With Phil's observation that it's a framework we're referencing rather than an organization I'm back to a simple text type for "educationalAuthority" though we may need some revision to the name and description to indicate that it's a framework, not an organization.

Regards to all -- especially Zoe for being so persistent.

Brandt

[Joshua Marks]

All,

 

Zoe also indicated that the Authority might be referred when no official “Text” (As Stewart uses the term) or reference framework exists (e.g. common/traditional use in a region.) This appears particularly so for grade level terms used in similar but not exactly the same way across systems and regions. It seems some flexible approach to defining the authoritative entity/organization is needed. So I concur with Brant that a simple text approach for the Authority is perhaps best. While less complete then Schema Organization proposal, it should work. Only vocabulary standards and practices can address the underlying terminology mess/diversity factor.

 

On related note I made earlier… We have an existing challenge identified where two (n) different promulgators and data representations of the a single standard framework might (do) exist. For example, we now have at least three data representations you might reference with a URI for the US K-12 common core. With the flexible approach, one might tag the authority as “CCSSO/NGA center” (The publishers of the standards), and reference any of the available URI resolution services/promulgators (The CCSSO’s canonical version or any other derivative there of). As the terms in the standard statement itself will be similar if not exactly the same (Even if the structure is changed), this approach could greatly improve the problem of “Cross walking” data from different alignment systems and different versions of the same standards framework (See the related Learning Registry discussion.)    

 

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.

 

From: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Brandt Redd
Sent: Monday, August 27, 2012 10:30 AM
To: lr...@googlegroups.com
Subject: Re: [LRMI] Zoe's proposal

 

I like where we're going with this. I agreed with Dan's suggestion that educationalAuthority could be an organization but I was concerned that the markup would get pretty complex with alignmentObject (an abstract type) incorporating Organization (another abstract type). This is logically really clean but the syntax of nested types may be beyond the average webmaster.

[Greg Grossmeier]

<quote name="Paul Libbrecht" date="2012-08-25" time="15:33:16 +0200">
> Le 25 ao�t 2012 � 06:45, Dan Brickley a �crit :

>
> > I'm fine with simple text labels here, but would also like to be
> > able to use any http://schema.org/Organization ( often but not
> > necessarily a http://schema.org/EducationalOrganization). This would
> > allow other attributes of the org (including identifiers) to be
> > included.
>
> Fully agree this would be helpful.

Agreed. I echo Brandt's concerns re complexity, but I don't see an
obvious way forward to address that.

> About the following: Is this agreed upon that educationLevel includes
> an entity with URL and text, just in case there are systems that
> already carry URL-based identifications of levels?

A URL in addition (or instead of) text is just fine.

[Greg Grossmeier]

Dan,

Other than the obvious "it'd be nice to have" reason for having the
value of educationalAuthority be a schema.org/Organization, is there a
specific use case you had in mind?

After re-reading this side of the sub-thread, I see the validity of
encouraging just the use of the text field.

Greg


<quote name="Joshua Marks" date="2012-08-27" time="11:18:24 -0700">

> All,
>
>
>
> Zoe also indicated that the Authority might be referred when no official
> "Text" (As Stewart uses the term) or reference framework exists (e.g.
> common/traditional use in a region.) This appears particularly so for grade
> level terms used in similar but not exactly the same way across systems and
> regions. It seems some flexible approach to defining the authoritative
> entity/organization is needed. So I concur with Brant that a simple text
> approach for the Authority is perhaps best. While less complete then Schema
> Organization proposal, it should work. Only vocabulary standards and
> practices can address the underlying terminology mess/diversity factor.
>
>
>

> On related note I made earlier. We have an existing challenge identified

> where two (n) different promulgators and data representations of the a
> single standard framework might (do) exist. For example, we now have at
> least three data representations you might reference with a URI for the US
> K-12 common core. With the flexible approach, one might tag the authority as
> "CCSSO/NGA center" (The publishers of the standards), and reference any of
> the available URI resolution services/promulgators (The CCSSO's canonical
> version or any other derivative there of). As the terms in the standard
> statement itself will be similar if not exactly the same (Even if the
> structure is changed), this approach could greatly improve the problem of
> "Cross walking" data from different alignment systems and different versions
> of the same standards framework (See the related Learning Registry
> discussion.)
>
>
>
> Joshua Marks
>
> CTO
>
> Curriki: The Global Education and Learning Community
>
> jma...@curriki.org
>
> www.curriki.org
>
>
>
>
>
> I welcome you to become a member

> <https://www.curriki.org/xwiki/bin/view/Main/JoinCurriki> of the Curriki
> community, to follow us on Twitter <http://twitter.com/Curriki> and to say
> hello on our blog <http://blog.curriki.org/> , Facebook
> <http://www.facebook.com/pages/Curriki/134427817464> and LinkedIn
> <http://www.linkedin.com/groupInvitation?groupID=1826931&sharedKey=277A033FE
> F70> communities.

> * "Northern Ireland"
> * "Royal Life Saving Society"
> * "North Carolina, State Board of Education, Department of
> Instruction"
> * "American Association for the Advancement of Science"

> <http://www.icbl.hw.ac.uk/~philb/> <http://www.icbl.hw.ac.uk/~philb/>

> Please note new email address: phil....@hw.ac.uk
>
>
>
> _____
>
>

> 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
> www.hw.ac.uk/researchleaders for further information and how to apply.
>
> Heriot-Watt University is a Scottish charity registered under charity number
> SC000278.
>
>
>

[Dan Brickley]

On Tuesday, 28 August 2012, Greg Grossmeier wrote:

Dan,

Other than the obvious "it'd be nice to have" reason for having the
value of educationalAuthority be a schema.org/Organization, is there a
specific use case you had in mind?

After re-reading this side of the sub-thread, I see the validity of
encouraging just the use of the text field.

You could encourage the text form, but the entity-based version gives more possibility for re-aggregating references to the same authority. I'd like at least to leave that door open. It also allows us to start thinking about properties of the authorities.... even if described elsewhere.

Dan

[Phil Barker]

On 27/08/12 18:30, Brandt Redd wrote:
> I like where we're going with this. I agreed with Dan's suggestion
> that educationalAuthority could be an organization but I was concerned
> that the markup would get pretty complex with alignmentObject (an
> abstract type) incorporating Organization (another abstract type).
> This is logically really clean but the syntax of nested types may be
> beyond the average webmaster.
>
> With Phil's observation that it's a framework we're referencing rather
> than an organization I'm back to a simple text type for
> "educationalAuthority" though we may need some revision to the name
> and description to indicate that it's a framework, not an organization.
>

+1 for a different name. educationalFramework seems the obvious choice

Phil

--
Ubuntu: not so much an operating system as a learning opportunity.
http://xkcd.com/456/

[Greg Grossmeier]

<quote name="Phil Barker" date="2012-08-28" time="08:30:14 +0100">

> +1 for a different name. educationalFramework seems the obvious choice

Thanks, Phil.

I had a draft email started yesterday with a new name and description,
but I must admit, I was stumped for a while and needed to walk away.

I like educationalFramework, would alignmentFramework be too generalized
(such that someone doesn't know what we're getting at?).

As for a strawman definition:
The framework which the other properties in this alignmentObject
instance are in reference to.

The "this alignmentObject instance" is different than any other
definition in Schema.or I've seen, but it seems useful to me.

Greg

[Dan Brickley]

On 28 August 2012 18:49, Greg Grossmeier <gr...@creativecommons.org> wrote:
> <quote name="Phil Barker" date="2012-08-28" time="08:30:14 +0100">
>> +1 for a different name. educationalFramework seems the obvious choice
>
> Thanks, Phil.
>
> I had a draft email started yesterday with a new name and description,
> but I must admit, I was stumped for a while and needed to walk away.
>
> I like educationalFramework, would alignmentFramework be too generalized
> (such that someone doesn't know what we're getting at?).
>
> As for a strawman definition:
> The framework which the other properties in this alignmentObject
> instance are in reference to.
>
> The "this alignmentObject instance" is different than any other
> definition in Schema.or I've seen, but it seems useful to me.

It's a little awkward but I think bearable. I'm +1 on a new name now
(prefering educationalFramework) as I realise the previous name made
me assume it would always be some kind of Organization.

Dan

[Greg Grossmeier]

<quote name="Dan Brickley" date="2012-08-28" time="19:18:58 +0200">

> >
> > As for a strawman definition:
> > The framework which the other properties in this alignmentObject
> > instance are in reference to.
>

> It's a little awkward but I think bearable. I'm +1 on a new name now
> (prefering educationalFramework) as I realise the previous name made
> me assume it would always be some kind of Organization.

Any other thoughts on this definition. I always assume my first pass
isn't good enough, and that was my first pass.

[Stuart Sutton]

A couple more stabs to prime the pump:

--The educationalFramework to which the resource being described is aligned.
--The educationalFramework to which the resource is aligned by the alignmentObject.

Stuart

[Greg Grossmeier]

Since no one else has responded...

I'll merge one of Stuart's and mine:
- The framework to which the resource being described is aligned.

Unless I hear otherwise, I think going with this is a good option.

Thanks all!

Greg

<quote name="Stuart Sutton" date="2012-08-28" time="11:28:42 -0700">

[Dan Brickley]

On Wednesday, 29 August 2012, Greg Grossmeier wrote:

Since no one else has responded...

I'll merge one of Stuart's and mine:
 - The framework to which the resource being described is aligned.

Unless I hear otherwise, I think going with this is a good option.

Thanks all!

Sounds good! Can you rework the schema.org submission and I'll get it queued up...?

Dan

[Greg Grossmeier]

<quote name="Dan Brickley" date="2012-08-29" time="22:25:22 +0200">

> Sounds good! Can you rework the schema.org submission and I'll get it
> queued up...?

Done: http://www.w3.org/wiki/WebSchemas/LearningResources#Vocabulary

[Joshua Marks]

Boy that is clumsy sounding grammar, but seems to say what it means.

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: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Greg
Grossmeier
Sent: Wednesday, August 29, 2012 1:19 PM
To: lr...@googlegroups.com
Subject: Re: [LRMI] Zoe's proposal

[Greg Grossmeier]

<quote name="Joshua Marks" date="2012-08-29" time="14:19:31 -0700">

> Boy that is clumsy sounding grammar, but seems to say what it means.

Yeah, all that indirectness. :/ Help appreciated.

[Monty Swiryn]

I, too, like where this is going.  Couple of remarks on the "grammar"...

In http://www.w3.org/wiki/WebSchemas/LearningResources#Vocabulary
you use "framework" to define "EducationalFramework".  (Weren't we taught in grade school not to use the term to describe the term? ;-)   Can you clarify, in the Description, what the definition of "framework" is?
Also, how about some examples?

Many thanks to all those who are making this happen.
Monty

[Monty Swiryn]

Question about the intention for ReadingLevel ...
Does the ReadingLevel construct take into consideration the ability to tag resources by different "types" of reading levels?  Here are some examples used by educational publishers (with whom I'm familiar) to provide educators targeted results in searching for specific products that meet their curriculum needs:
Guided Reading Level: values typically assigned on a scale from "A" to "Z"
Early Intervention Level: values typically assigned on a scale from "1" to "30"
Interest Level: values are similar to "grade levels" but the definition is obviously different
Lexile Level: check it out here: http://lexile.com/
Developmental Reading Assessment Level: values typically range from "1" to "44"
EDL Level (For Products in Spanish): values typically range from "1" to "38" (http://www.cal.org/twi/assessments.pdf)

In other words, how would a publisher indicate that a product/resource fits a "reading level name/type" and has a value of "level"?

Examples would be most appreciated.
thx
Monty

[Monty Swiryn]

Hi Greg,

Here's an example page:

http://sundancepub.com/product.html?record@S8505

thx

monty

On 9/5/2012 9:31 AM, Greg Grossmeier wrote:

Hi Monty,

That's a good question. Do you have an example page that has that
information on it; either one or more reading level type attributes?
I'll take a swing at marking it up if you can give me one.

Best,

Greg

 

[Greg Grossmeier]

<quote name="Monty Swiryn" date="2012-09-05" time="11:26:01 -0700">

> Here's an example page:
> http://sundancepub.com/product.html?record@S8505

Thanks for that, Monty.

Here's my marked up version:
http://grossmeier.net/files/tmp/Laugh-Out-Loud_Chapter_Books.html

I just did the first book listed in that table.


Here is what the Google Rich Snippet Testing Tool pulls out:
http://goo.gl/TfUeo

Reproduced here:

Item
Type: http://schema.org/book
name
text = The Adventures of Zip Velocity
href = http://sundancepub.com/c/@FIYF3g691ZhxA/Pages/product.html?nocache@1+record@P8532
educationalalignment = Item( 1 )
educationalalignment = Item( 2 )
educationalalignment = Item( 3 )
numberofpages = 32

Item 1
Type: http://schema.org/alignmentobject
alignmentype = readingLevel
targetname = K

Item 2
Type: http://schema.org/alignmentobject
alignmentype = textComplexity
targetname = 600L

Item 3
Type: http://schema.org/alignmentobject
alignmentype = educationLevel
targetname = 2 - 6



Pertinent part of the HTML
--------------------------

(slightly reformatted in this email, content is the same)


<tr itemscope itemtype="http://schema.org/Book">

<td>
<a href="http://sundancepub.com/c/@FIYF3g691ZhxA/Pages/product.html?nocache@1+record@P8532"
itemprop="name">The Adventures of Zip Velocity</a>
<img src="Laugh-Out-Loud_Chapter_Books_files/new_sm.gif" align="absmiddle">
</td>

<td itemprop="educationalAlignment" itemscope
itemtype="http://schema.org/AlignmentObject" align="center">
<meta itemprop="alignmentype" content="readingLevel">
<span itemprop="targetName">K</span>
</td>

<td itemprop="educationalAlignment" itemscope
itemtype="http://schema.org/AlignmentObject" align="center">
<meta itemprop="alignmentype" content="textComplexity">
<span itemprop="targetName">600L</span>

<td itemprop="educationalAlignment" itemscope
itemtype="http://schema.org/AlignmentObject" align="center">
<meta itemprop="alignmentype" content="educationLevel">
<span itemprop="targetName">2 - 6</span>
</td>

<td align="center" itemprop="numberOfPages">32
</td>

</tr>


Best,

Greg


<quote name="Monty Swiryn" date="2012-09-05" time="11:26:01 -0700">

[Monty Swiryn]

Hi Greg,

Thanks for doing this. This is beginning to become much clearer.  It gives a good idea about the complexity of marking up products with these different types of educational attributes.  I hope this makes things clearer for others as well.

I have some questions about your markup...

1. The top level schema for this product would be "Thing > CreativeWork > Book", correct?

Should it also be tagged as "Thing > Product", since it's something for sale?

2. "schema.org/AlignmentObject" doesn't go to a page in schema.org.  Is that because LRMI hasn't been included in schema.org yet?

Are you referring to http://www.lrmi.net/the-specification/alignment-object ?

3. Can a publisher use any value for "alignmentype"?  Is this the "language" issue that has been discussed in this forum?

For example:

   3.1.  "Reading Level" is not necessarily the same as "Guided Reading Level".  Some products will have both, and the values might be different. What would the markup be if we needed to use both of these?

  3.2. "Text complexity" does not specifically state "Lexile Level", which is what some teachers will be using in their searches.

  3.3  "educationLevel" -- This may not be descriptive enough for the audience of educators who may be interested in this book.  They need to be able to see  terms that they commonly use, such as "Interest Level" vs "Grade Level" vs "Age Range" vs all those different scales of other "reading" levels.

So do we just add tags for all these niche-specific characteristics?

4. I don't see the  itemprop "numberOfPages" in schema.org for Book, nor in the LRMI spec.  Am I missing something?  Does this mean we can add our own "itemprop" types?  If so, it might be problematic for itemprops that are not clearly specified somewhere.  Publishers might use different terns for this, eg: "numberOfPages", "Pages", "Length", etc.  How will the publishers agree on the term to use in cases like this?

5. If we're going to use terms such as these for this product, do we need to add tags for identifying the "framework" source, as discussed regarding Zoe's concerns?

Many thanks again,

Monty

[Greg Grossmeier]

Hi Monty,

I'm on a sick day today (ongoing migraine) but as I was in the shower I
realized I didn't do an obvious bit of markup:

<quote name="Monty Swiryn" date="2012-09-06" time="13:13:04 -0700">

> 5. If we're going to use terms such as these for this product, do we need
> to add tags for identifying the "framework" source, as discussed regarding
> Zoe's concerns?

Yes, I'll add that tomorrow, and respond to your other points.

[Joshua Marks]

The cause of your pain is not LRMI specs, I hope Greg?

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: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Greg
Grossmeier
Sent: Thursday, September 06, 2012 3:13 PM
To: lr...@googlegroups.com

[Monty Swiryn]

Many thanks, Greg.  Hope you feel better.

Monty

[Greg Grossmeier]

Hello Monty,

Updated example at the same url:
http://grossmeier.net/files/tmp/Laugh-Out-Loud_Chapter_Books.html

<quote name="Monty Swiryn" date="2012-09-06" time="13:13:04 -0700">

> 1. The top level schema for this product would be "Thing > CreativeWork >
> Book", correct?

Correct.

> Should it also be tagged as "Thing > Product", since it's something for
> sale?

You can use the "offer" field from Book, and fill it with information
from the intangible "Offer" (this sentence is case sensitive).

The example markup on http://schema.org/Book shows how this works.

> 2. "schema.org/AlignmentObject" <http://schema.org/AlignmentObject> doesn't

> go to a page in schema.org. Is that because LRMI hasn't been included in
> schema.org yet?

Right.

> Are you referring to http://www.lrmi.net/the-specification/alignment-object
> ?

That page is now out of date (I am holding off on updating it until the
final version is included in Schema.org to prevent unneeded confusion).
You can see the latest development version here:
http://www.w3.org/wiki/WebSchemas/LearningResources

> 3. Can a publisher use any value for "alignmentype"? Is this the
> "language" issue that has been discussed in this forum?
> For example:
> 3.1. "Reading Level" is not necessarily the same as "Guided Reading
> Level". Some products will have both, and the values might be different.
> What would the markup be if we needed to use both of these?

This need is addressed by "educationalFramework." It tells you what
framework to interpret the alignment within. See the updated example.

> 3.2. "Text complexity" does not specifically state "Lexile Level", which
> is what some teachers will be using in their searches.

Again, this is addressed by "educationalFramework."

> 3.3 "educationLevel" -- This may not be descriptive enough for the
> audience of educators who may be interested in this book. They need to be
> able to see terms that they commonly use, such as "Interest Level" vs
> "Grade Level" vs "Age Range" vs all those different scales of other
> "reading" levels.
> So do we just add tags for all these niche-specific characteristics?

You are free to use whatever new terms you feel are needed in the
alignmentType field.

However, just a reminder, typicalAgeRange is already included in LRMI.

> 4. I don't see the itemprop "numberOfPages" in schema.org for Book, nor in
> the LRMI spec. Am I missing something?

You missed it on the Book page. Scroll to the bottom of the list of
terms. It is the last one listed (before the examples start).

> 5. If we're going to use terms such as these for this product, do we need
> to add tags for identifying the "framework" source, as discussed regarding
> Zoe's concerns?

The example is updated to use that term. One relevant section reproduced
below:

<td itemprop="educationalAlignment"
itemscope itemtype="http://schema.org/AlignmentObject" align="center">
<meta itemprop="alignmentype" content="textComplexity">

<meta itemprop="educationalFramework" content="Lexile">

<span itemprop="targetName">600L</span>

</td>

[Greg Grossmeier]

I should add a general suggestion to everyone planning on implementing
LRMI:

First, you should think to yourself: "I'm going to implement Schema.org
support on my site."

After you have marked up everything you can with what is in Schema.org,
then begin to add in the extra LRMI fields.

This will clear up some of the confusion expressed below, and, it is
also the right way to go about it.

Best,

Greg

<quote name="Greg Grossmeier" date="2012-09-07" time="09:30:40 -0700">

[Monty Swiryn]

Hi Greg,

Most helpful.  Thanks for taking the time to answer all my "dumb" questions.

Just curious...  has anyone else had similar questions/confusion?  Have Greg's answers help to clear them up?  It has for me in a BIG way!

Now I have to work on getting these marked-up pages generated from the product database, as we currently do.  Certainly don't want to markup thousands of pages by hand ;)  I guess this means we won't be using the Tagger app, correct?

I'd love to hear from publishers about whether or not they plan to use the Tagger, and how they plan to integrate it into their current website systems.

thx
monty

[Joshua Marks]

Monty,

 

We have been trying to get more detailed and varied examples, which will exercise the flexibility and clarity of the spec. There will be more example soon, I expect, and evolving practices based on them.

 

Regarding a tagging tool- If you already have the metadata in your resource database AND you can generate new pages to contain the tags, I see no benefit to using a tagging tool. We will be doing this in Curriki soon as well. If you want to change or add metadata, that might be a different story. Where a tagger is helpful is getting all sources to use a common vocabulary and set of frameworks. Where it is not so helpful is actually getting that data and those tags into your pages.  

 

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.

 

From: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Monty Swiryn
Sent: Friday, September 07, 2012 10:25 AM
To: lr...@googlegroups.com
Subject: Re: Example Markup - alignmentType=readingLevel/textComplexity/educationalLevel

 

Hi Greg,

[Monty Swiryn]

Thanks, Joshua,

So does this imply that the tagging tool will produce the "general" tags as Greg described? 
Ex: It will first put in all the the basic schema.org tags, and then the LRMI tags?
This would be a great use of the Tagger... We could use it to basically generate a template that could be used/extended for all products in the database.

thx
monty

[Monty Swiryn]

Hi Greg,

Just to clarify... Does the sample page you marked up for me include all the recommended schema.org tags?

thx
monty

[Greg Grossmeier]

<quote name="Monty Swiryn" date="2012-09-07" time="11:27:49 -0700">

> Just to clarify... Does the sample page you marked up for me include all
> the recommended schema.org tags?

I didn't go through and mark up the entire page, no. I was responding to
the need for an example to address the question:

"Does the ReadingLevel construct take into consideration the ability to
tag resources by different "types" of reading levels?"

So, I just marked up the first line of the table on that page where the
reading level information is located.

That page should also be marked up with description, title/name, author
(if appropriate), offer/price, audience/typicalAgeRange (if applicable),
awards, copyrightHolder, useRightsUrl, datePublished, genre, inLanguage,
provider/publisher, thumbnailUrl, bookFormat, isbn... (as appropriate,
of course).

:)

All of those are described on http://schema.org/Book

[Greg Grossmeier]

I'll let those who are working on the Tagger Tool comment on this as I
haven't been following it as closely as I should to answer
knowledgeably.

Greg


<quote name="Monty Swiryn" date="2012-09-07" time="11:24:25 -0700">

[Monty Swiryn]

Got it!
Thanks again, Greg.

Monty

[Greg Grossmeier]

Hello all (especially Dan),

<quote name="Dan Brickley" date="2012-08-25" time="06:45:20 +0200">
> > Term: educationalAuthority
> >
> > Definition: the authority which defines the other attributes in
> > alignmentObject
> >
> > Example: Royal Life Saving Society, PRINCE2, Welsh national
> > curriculum
> >
> > Because: the other descriptive statements are only meaningful in the
> > context of the defining body
>
> I'm fine with simple text labels here, but would also like to be able
> to use any http://schema.org/Organization ( often but not necessarily
> a http://schema.org/EducationalOrganization). This would allow other
> attributes of the org (including identifiers) to be included.

To do our best in making sure that LRMI is usable for machines we should
try to eliminate any special cases. Having the expected type for this
property be Text or Organization creates such a special case any
developer trying to use LRMI will run into.

Why? schema.org/Text and schema.org/Organization have different roots
(and indeed don't even share a root). The root of Text is DataType and
the root of Organization is Thing.

Thus, I recommend the use of *just* Organization.

This makes the markup of the name of the educationalAuthority just a
tiny bit more complex (just another <span>), but in the end, more
accurate and usable for machines.

I chose to use Organization over just Text as I couldn't think of an
educationalAurhority that doesn't fit within
http://schema.org/Organization (or the subtypes like
EducationalOrganization, GovernmentOrganization, etc).

Unless someone strongly objects, I'll make this change on Tuesday the
25th.

Greg

[Greg Grossmeier]

<quote name="Greg Grossmeier" date="2012-09-21" time="16:12:11 -0700">

> This makes the markup of the name of the educationalAuthority just a
> tiny bit more complex (just another <span>), but in the end, more
> accurate and usable for machines.

I forgot to add that having expectedTypes be of differing roots appears
to be not an option in Schema.org (for good reasons). No other term that
I could find allows it. With a well structured ontology that is the
right way of doing it, in fact.

Have a good weekend,

[Brandt Redd]

I disagree, in fact, somehow I missed the switch from "educationalFramework" to "educationalAuthority". I continue to favor the former term.

A concrete example to illustrate: This week I met with people doing LRMI tagging -- Mark Luetzelschwab from Agilix building an LRMI tagging tool under contract with the Shared Learning Collaborative (SLC). They've been coordinating with the Michael Jay and the AEP to ensure that the tagging tool works equally well for publishers and the SLC states.

The problem is that the publishers and teachers want to align content to US Grade levels. They had been doing this by mapping grades to ages and using the "typicalAgeRange" property. I explained that "typicalAgeRange" is intended to reflect content interest (e.g. cute animals vs. vampire novels vs. literature) and that the proper way to do this is with educationalAlignment.

This shouldn't be a big surprise, it was a similar issue with the grade-level equivalents in the UK that caused Zoe to bring this to our attention.

Anyway, there is no organization that establishes the K-12 grade system in the US. It's simply a convention -- an informal agreement between the 50 state education agencies. If constrained to an organization then we would have 50 different organizations making it difficult to correlate between states. And we can't cite the US department of education without freaking out a bunch of people who will say that we're establishing a national curriculum.

So, I can tag "educationalFramework: US K-12". In this case, the framework is a convention -- an informal agreement -- not an organization. But if I have to name an organization, there's not one to cite.

So, I recommend keeping "educationalFramework" (which is in the current W3C submission at http://www.w3.org/wiki/WebSchemas/LearningResources) and leaving the value as "text". I see no advantage to changing it to "educationalAuthority" and "organization" and I see a lot of disadvantages.

Thanks,

Brandt

[Brandt Redd]

Another objection. What happens when the same organization publishes more than one framework? For example, Metametrics publishes both Lexile (for text complexity) and Quantile (for mathematics progressions).

[Joshua Marks]

Two different frameworks. The Authority is useful to link AB, ASN’s and the CCSSO’s different expressions of the same framework, for example.

 

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.

 

From: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Brandt Redd

Sent: Friday, September 21, 2012 6:05 PM
To: lr...@googlegroups.com

[Joshua Marks]

Hmmm, convention vs. organization… We did discuss that too, and the educationalAutority element that attributes the educaitonalFramework to some Organization (Or maybe text) is not required for the alignment to a leveling framework. If there is no framework to align to specifically, there is no alignment. If it is truly by convention or tradition, someone must still express that convention as a framework. This suggests the Authority might be “US convention”. However the CommonCore itself is (or contains) a leveling framework, so would that not the be authority in this case?

 

Joshua Marks

CTO

Curriki: The Global Education and Learning Community

jma...@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.

 

From: lr...@googlegroups.com [mailto:lr...@googlegroups.com] On Behalf Of Brandt Redd
Sent: Friday, September 21, 2012 5:44 PM
To: lr...@googlegroups.com
Subject: Re: Text vs Organization for educationalAuthority

 

I disagree, in fact, somehow I missed the switch from "educationalFramework" to "educationalAuthority". I continue to favor the former term.

[Brandt Redd]

I really hesitate to write the following because I'm anxious to resolve this rather than bring up more questions but, "fools rush in...":

When we originally conceived of AlignmentObject, we designed it to be compatible with any alignment between some object and some taxonomy. The term "education" is only added at the in the property name -- "educationalAlignment." The object type was designed to be universal.

So, naming the a property of "AlignmentObject," to "educationalAlignment" or "educationalAuthority" constrains this object type to the education domain where previously that wasn't the case. Just because the only use of the object type is presently in education doesn't mean it will remain so.

Apologies,

Brandt

[Greg Grossmeier]

<quote name="Brandt Redd" date="2012-09-21" time="17:43:45 -0700">

> I disagree, in fact, somehow I missed the switch from
> "educationalFramework" to "educationalAuthority". I continue to favor the
> former term.

I think I messed up there. I knew something was off.

You're right, educationalFramework is the agreed upon term. I messed up
do to my quoting of Dan's email which quoted a previous version (it's
just that this was the email in which the Text vs Organization issue
came up).

My apologies for that confusion, everyon.

If you are unsure, check
http://www.w3.org/wiki/WebSchemas/LearningResources for the latest
version, that is the canonical source right now (pending Schema.org
integration and then lrmi.net will be updated).

[snip a useful explanation of why you make this recommendation]

> So, I recommend keeping "educationalFramework" (which is in the current W3C
> submission at http://www.w3.org/wiki/WebSchemas/LearningResources) and
> leaving the value as "text". I see no advantage to changing it to
> "educationalAuthority" and "organization" and I see a lot of disadvantages.

That is a solid recommendation. And it was the first one we proposed.

Given this strong feedback, and my initial confusion (which colored my
recommendation, unfortunately), I would like to agree with Brandt on
this.

I'll let it sit for the weekend (I almost hesitate to send this after
just doing the dishes and sitting down for 2 minutes, but... fools rush
in as Brandt says), though.

[Greg McFall]

Sorry for joining this conversation so late, but in my opinion neither "Text" nor "Organization" is appropriate as a data type for the educationalFramework property.  Text is a primitive data type and an educational framework clearly is NOT a primitive.  

It seems to me that we ought to define EducationalFramework as a new class and use it as the type for the educationalFramework property.

Some organizations may be declared to be an EducationalFramework (in addition to being an Organization), but these are fundamentally different concepts.  There is no rule which says that a given object must belong to only one class.

Also, I think that "Text" is the wrong data type for "alignmentType".  I would prefer that we follow an approach similar to the medical community for how enumerations are handled.  See MedicalEnumeration.  Following this design pattern, we should define a new class for each kind of enumeration.  Thus, AlignmentType should be declared as a class, and then practitioners would define instances of that class.  The LRMI standard could define a core vocabulary of instances of AlignmentType (named 'assessed', 'teaches', 'requires', 'textComplexity', 'readingLevel', 'educationalSubject', and 'educationLevel'), but communities of practice are free to define other instances.  The vocabulary is open-ended.

[Jim Goodell]

Let's not forget the goal, that publishers USE LRMI to tag content. What is ideal for the consuming machine or search engine (complex types) is at odds with what is best to maximize use by content publishers.

[Greg Grossmeier]

<quote name="Jim Goodell" date="2012-09-24" time="07:47:02 -0700">

Good point, Jim.

I think adding extra complexity along the lines of yet another type
(thus requiring yet more nested <span>'s) will negatively effect LRMI's
adoption.

Thanks for the feedback, everyone.

[Greg McFall]

I don't really see the conflict.  Effectively, what I am suggesting is that the values should be URIs that may (or may not) point to resources instead of plain text.  Just because we introduce a new type like EducationalFramework does not necessarily mean that the data values must be encapsulated using nested spans.  See the discussion about canonical references using <link> at  http://schema.org/docs/gs.html#advanced_enum.

In my opinion, <link> with a URI value is hardly more complex than using a plain text value, but the result is much richer semantically.  So I don't really see how this would be a burden on adoption.  Furthermore, I don't believe the goal is simply to drive adoption.  If the resulting data is not useful by consuming machines then what's the point?

[Phil Barker]

On 26/09/2012 08:53, Greg McFall wrote:

I don't really see the conflict.  Effectively, what I am suggesting is that the values should be URIs that may (or may not) point to resources instead of plain text.  Just because we introduce a new type like EducationalFramework does not necessarily mean that the data values must be encapsulated using nested spans.  See the discussion about canonical references using <link> at  http://schema.org/docs/gs.html#advanced_enum.

hmm, <link>s in the <body> of the document. I've done that few times myself but always worried about whether it is valid HTML.  Anyway...

I don't see how a link nested in a span is simpler than a span nested in a span. It's not just a complexity problem: when you nest schema elements you put constraints on what order information should be presented in an HTML page. 

That example illustrates another "feature" of using complex types: they get ignored. Author should be a Person object not text. I imagine that the same will happen if we make EducationalFramework a complex object: if webmasters can ignore that it is an object and just provide a text value they will. This is a strength and a weakness of schema: I don't know whether it means that we can go ahead and make objects galore, knowing that people will ignore them if they like and Google will cope (but remember, we're already one object down from the resource being described and you can only flatten so much), but I suspect that the more complex the spec looks the less likely people will use it so I would prefer not to.

I think that if you want a perfect model for learning resource metadata you might be better using RDF rather than trying mark up the semantics of what is put on web pages.

What I would really like to see is a credible use case for this being necessary in order to make a major improvement in web-scale search for learning materials--I mean something that Google might adopt. For the sake of adoption by content providers this must be focussed on providing sematic mark-up for information that is already exposed on the web, or at least is already stored or implicit in the back-end data, without placing too many constraints on the appearance of the webpages.

I am naturally drawn to providing URIs for everything, so have sympathy for this proposal; but my head says "not in version 1" for LRMI.

Regards, Phil

In my opinion, <link> with a URI value is hardly more complex than using a plain text value, but the result is much richer semantically.  So I don't really see how this would be a burden on adoption.  Furthermore, I don't believe the goal is simply to drive adoption.  If the resulting data is not useful by consuming machines then what's the point?

On Monday, September 24, 2012 6:14:25 PM UTC-4, Greg Grossmeier wrote:

<quote name="Jim Goodell" date="2012-09-24" time="07:47:02 -0700">
> Let's not forget the goal, that publishers USE LRMI to tag content.
> What is ideal for the consuming machine or search engine (complex
> types) is at odds with what is best to maximize use by content
> publishers.

Good point, Jim.

I think adding extra complexity along the lines of yet another type
(thus requiring yet more nested <span>'s) will negatively effect LRMI's
adoption.

Thanks for the feedback, everyone.

Greg


--
Greg Grossmeier
Education Technology & Policy Coordinator
twitter: @g_gerg / identi.ca: @greg / skype: greg.grossmeier

-- 
<http://www.icbl.hw.ac.uk/~philb/>

---

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 www.hw.ac.uk/researchleaders for further information and how to apply.

Heriot-Watt University is a Scottish charity registered under charity number SC000278.

[Monty Swiryn]

Question... has anyone done, or is planning to do, a survey of educational publishers regarding these issues?  It would seem valuable to get the opinions of publishers to see how helpful they think these ideas will be to their tagging efforts. 

Note that many publishers, particularly those with hundreds or thousands of products in a database, will not be using a tagging tool.  They may have something to say about the amount of work involved and/or the appropriateness of these options to their educational resources.

thx
Monty

[Greg McFall]

On Wednesday, September 26, 2012 5:41:25 AM UTC-4, Phil Barker wrote:

I don't see how a link nested in a span is simpler than a span nested in a span. It's not just a complexity problem: when you nest schema elements you put constraints on what order information should be presented in an HTML page. 

The use of <link> is simpler exactly because it allows you to break out of the "constraints on what order information should be presented in an HTML page".  The <link> element can be placed anywhere inside the parent span because it has no visual presentation.  It allows you to say that one resource is associated with some other resource without having to display any visible elements, and hence it could go anywhere.  Of course, if you want to provide a hyperlink to the related resource, you could use an anchor tag instead of a <link>.  And if you want to embed a complete description of the related resource, you can do that too by nested more spans within spans.

 

That example illustrates another "feature" of using complex types: they get ignored. Author should be a Person object not text. I imagine that the same will happen if we make EducationalFramework a complex object: if webmasters can ignore that it is an object and just provide a text value they will. This is a strength and a weakness of schema: I don't know whether it means that we can go ahead and make objects galore, knowing that people will ignore them if they like and Google will cope (but remember, we're already one object down from the resource being described and you can only flatten so much),

 

I am inclined to paraphrase your statement:

If webmasters can ignore that it is an object and just provide a URI value they will.

Any place a webmaster can provide plain text, they should be able to provide a URI in a <link>.  I really don't see how using a URI is an inhibitor to adoption. Maybe you are worried that the URI must resolve to a webpage or some other resource that can be downloaded.  But, in fact, that is not necessary.  A URI need not resolve to anything; it can be nothing more than a somewhat formal name for a Thing.  It is a name that has the benefit of being globally unique.

but I suspect that the more complex the spec looks the less likely people will use it so I would prefer not to.

I totally agree with this statement.  Perhaps we should define different levels of sophistication in a manner similar to the way that OWL defines different levels: OWL Lite, OWL DL, OWL Full.

 

I think that if you want a perfect model for learning resource metadata you might be better using RDF rather than trying mark up the semantics of what is put on web pages.

Yes, I absolutely want to use RDF.  But I want the microdata markup to be consistent with the RDF ontology. Indeed, I am in the process of defining an RDF Ontology for LRMI, and that's why I am struggling with this issue.  

What I would really like to see is a credible use case for this being necessary in order to make a major improvement in web-scale search for learning materials--I mean something that Google might adopt. For the sake of adoption by content providers this must be focussed on providing sematic mark-up for information that is already exposed on the web, or at least is already stored or implicit in the back-end data, without placing too many constraints on the appearance of the webpages.

I think we miss an opportunity if we limit the problem space exclusively to search by humans.  As a software architect for Pearson Education, I want our systems to work with learning resources that our (not Google) search engines discover.  We want to empower users to create custom products that blend our publisher-provided resources with free and open educational resources.  But the cost of assembling custom products is very high if we need to discover resources through one channel (search), but then build a separate feed for the "back-end data" or define a process to "scrape-and-clean" the microdata.

 

I am naturally drawn to providing URIs for everything, so have sympathy for this proposal; but my head says "not in version 1" for LRMI.

I totally get this point.  But we should be careful to ensure forward compatibility with later versions that provide richer semantics.  To that end, maybe we should define the type of "educationalFramework" to be "schema:Thing" in version 1.  A Thing can be either Text or an object, and (in version 1) we make no commitment to what kind of objects are suitable.

[Phil Barker]

On 28/09/2012 17:22, Greg McFall wrote:

On Wednesday, September 26, 2012 5:41:25 AM UTC-4, Phil Barker wrote:

I don't see how a link nested in a span is simpler than a span nested in a span. It's not just a complexity problem: when you nest schema elements you put constraints on what order information should be presented in an HTML page. 

The use of <link> is simpler exactly because it allows you to break out of the "constraints on what order information should be presented in an HTML page".  The <link> element can be placed anywhere inside the parent span because it has no visual presentation.  It allows you to say that one resource is associated with some other resource without having to display any visible elements, and hence it could go anywhere.

yes, that's true, being invisible helps.

 Of course, if you want to provide a hyperlink to the related resource, you could use an anchor tag instead of a <link>.  And if you want to embed a complete description of the related resource, you can do that too by nested more spans within spans.

 

That example illustrates another "feature" of using complex types: they get ignored. Author should be a Person object not text. I imagine that the same will happen if we make EducationalFramework a complex object: if webmasters can ignore that it is an object and just provide a text value they will. This is a strength and a weakness of schema: I don't know whether it means that we can go ahead and make objects galore, knowing that people will ignore them if they like and Google will cope (but remember, we're already one object down from the resource being described and you can only flatten so much),

 

I am inclined to paraphrase your statement:

If webmasters can ignore that it is an object and just provide a URI value they will.

That's not a paraphrase of my statement, and it is not what I think would happen: I really don't see many people using the complex type for subject in the about property, or publisher and I see them using plain text strings, not URIs. I guess we know different webmasters.

Any place a webmaster can provide plain text, they should be able to provide a URI in a <link>.  I really don't see how using a URI is an inhibitor to adoption. Maybe you are worried that the URI must resolve to a webpage or some other resource that can be downloaded.  But, in fact, that is not necessary.  A URI need not resolve to anything; it can be nothing more than a somewhat formal name for a Thing.  It is a name that has the benefit of being globally unique.

I know that, you know that. Most of the webmasters I know still won't use URLs that don't resolve to something :(

 

I am naturally drawn to providing URIs for everything, so have sympathy for this proposal; but my head says "not in version 1" for LRMI.

I totally get this point.  But we should be careful to ensure forward compatibility with later versions that provide richer semantics.  To that end, maybe we should define the type of "educationalFramework" to be "schema:Thing" in version 1.  A Thing can be either Text or an object, and (in version 1) we make no commitment to what kind of objects are suitable.

A compromise: do as schema has done for author, about. Make educational authority a Thing, so that people who want to provide URLs have the means to do so, but keep the examples simple by only providing a text value. 

Phil

[Brandt Redd]

To me the key factor is a point Phil made earlier in this thread. The major search engines pretty much ignore things that are invisible. Otherwise they are vulnerable to index manipulation. Also, the search engines are not going to dereference URLs and incorporate metadata at the other end. That's why AlignmentObject includes targetName and targetDescription. These are denormalizations to support the search engine use case.

Regards,

Brandt

[Greg McFall]

The schema.org site explicitly endorses the use use of "invisible" <link> elements in conjunction with microdata. See http://schema.org/docs/gs.html#advanced_enum.  Since schema.org was created as a collaborative effort of the major search providers (Bing, Google, Yahoo! and Yandex), I would think this means that they are not ignoring microdata markup in <link> elements.

[Greg Grossmeier]

Hi Greg,

<quote name="Greg McFall" date="2012-10-04" time="19:07:01 -0700">

> The schema.org site explicitly endorses the use use of "invisible" <link>
> elements in conjunction with microdata. See
> http://schema.org/docs/gs.html#advanced_enum. Since schema.org was created
> as a collaborative effort of the major search providers (Bing, Google,
> Yahoo! and Yandex), I would think this means that they are not ignoring
> microdata markup in <link> elements.

That is part of the story.

What each of the search engines DO with that data is the question.
They're all consuming it, there's no real pre-processing, that's too
expensive, but which parts they choose to display and/or make decisions
based off of is the kicker.

Google, specifically, has a very tenuous relationship with metadata.
Metadata can be spoofed in the same way people spoof the keyword class
in HTML, or what people did back in the 90s (and a little beyond) where
the "footer" of a webpage included about 100 different, not all related,
words.

What I'm trying to say is: Yes, they'll consume it. And, if (and only
if, probably) they can be relatively sure that your metadata is
conforming to best practices of non-spammers/high quality content then
they'll do something with it.

How can you make sure you follow best practices of non-spammers/high
quality content producers? Well, we're doing a bit of that here, having
these discussions that will be the best practices of our community.
Hopefully we get a critical mass of people who will implement
LRMI/Schema.org in a similar fashion so that Google et al see it as best
practices.

That's the goal, really...

[Monty Swiryn]

Hi Greg,

I was trying to look at the example you created, but it appears that the URL doesn't work anymore.
Do you still have it available on line?

thx
monty

On Friday, September 7, 2012 9:31:32 AM UTC-7, Greg Grossmeier wrote:

Hello Monty,

Updated example at the same url:
http://grossmeier.net/files/tmp/Laugh-Out-Loud_Chapter_Books.html

<quote name="Monty Swiryn" date="2012-09-06" time="13:13:04 -0700">
> 1. The top level schema for this product would be "Thing > CreativeWork >
> Book", correct?

Correct.

> Should it also be tagged as "Thing > Product", since it's something for
> sale?

You can use the "offer" field from Book, and fill it with information
from the intangible "Offer" (this sentence is case sensitive).

The example markup on http://schema.org/Book shows how this works.

> 2. "schema.org/AlignmentObject" <http://schema.org/AlignmentObject> doesn't
> go to a page in schema.org.  Is that because LRMI hasn't been included in
> schema.org yet?

Right.

> Are you referring to http://www.lrmi.net/the-specification/alignment-object
> ?

That page is now out of date (I am holding off on updating it until the
final version is included in Schema.org to prevent unneeded confusion).
You can see the latest development version here:
http://www.w3.org/wiki/WebSchemas/LearningResources

> 3. Can a publisher use any value for "alignmentype"?  Is this the
> "language" issue that has been discussed in this forum?
> For example:
>    3.1.  "Reading Level" is not necessarily the same as "Guided Reading
> Level".  Some products will have both, and the values might be different.
> What would the markup be if we needed to use both of these?

This need is addressed by "educationalFramework." It tells you what
framework to interpret the alignment within. See the updated example.

>   3.2. "Text complexity" does not specifically state "Lexile Level", which
> is what some teachers will be using in their searches.

Again, this is addressed by "educationalFramework."

>   3.3  "educationLevel" -- This may not be descriptive enough for the
> audience of educators who may be interested in this book.  They need to be
> able to see  terms that they commonly use, such as "Interest Level" vs
> "Grade Level" vs "Age Range" vs all those different scales of other
> "reading" levels.
> So do we just add tags for all these niche-specific characteristics?

You are free to use whatever new terms you feel are needed in the
alignmentType field.

However, just a reminder, typicalAgeRange is already included in LRMI.

> 4. I don't see the  itemprop "numberOfPages" in schema.org for Book, nor in
> the LRMI spec.  Am I missing something?  

You missed it on the Book page. Scroll to the bottom of the list of
terms. It is the last one listed (before the examples start).

> 5. If we're going to use terms such as these for this product, do we need
> to add tags for identifying the "framework" source, as discussed regarding
> Zoe's concerns?

The example is updated to use that term. One relevant section reproduced
below:
 
<td itemprop="educationalAlignment"
itemscope itemtype="http://schema.org/AlignmentObject" align="center">
  <meta itemprop="alignmentype" content="textComplexity">
  <meta itemprop="educationalFramework" content="Lexile">
  <span itemprop="targetName">600L</span>
</td>

Best,

[Monty Swiryn]

HI Greg,
Just checking back in now that the LRMI is official.  Congrats to everyone involved.

I'm trying to get to the sample page you marked up:
http://grossmeier.net/files/tmp/Laugh-Out-Loud_Chapter_Books.html
but the  link doesn't work.
Do you happen to still have it somewhere?

much appreciated.
Monty

--------------------