Status of inititiative ?

[gdupont]

Hi all,

It's long time no one send anything on this group... I wonder what
happened ? Does everybody let APML down for other activities or is
there another initiative that has taken APML place ?

I have few idea in mind to start again thinking on APML, but I want to
know if I'm not alone before...

[Chris Saad]

I think APML is being used in all sorts of contexts. I'm not aware of anything else that would do the job. In fact I keep finding new use cases for it as I talk to people.

Re: Status - I think we were VERY close to ratifying v1 - Paul what open issues did we have around that?

Chris

Chris Saad
VP Product and Community Strategy
JS-Kit creators of Echo

Google Talk: chris...@gmail.com Skype: chris.saad

Paying Attention? What is Echo?

[Paul Jones]

I believe the key issue that needs to be resolved is as to how best to strike the balance between making the data easy to produce, and making the data useful to consume. As far as I can tell, the (main) possible paths from here are:

  - Push forward with as simple a specification as possible based on the "iterative" changes from the pre-1.0 APML specifications;

  - Produce an RDF-style grammar that breaks from the stylings of the pre-1.0 specifications, but more closely mirrors the formats dealt with by many semantic applications.

I'd probably assert that the two options above express the ends of the spectrum that I mentioned - the key blocker is finding a way to strike a balance! As always, more possible use cases will always help in clarifying the best decision to make here, so please feel free to bring them on.

Paul.

[thieme]

hi all,
Recently, I started a PhD on networked learning and open educational resources. The focus I want to bring to the research is on user-generated metadata; making profiles on people and on content by looking at the usage & interaction and relating that to value (within a context). Eg; a relatively large number of comments on a post means that the "discussion value" is probably high. If that's considered important within a community, than it means that the author has "value". Suppose this author has 10 more contributions to a community, and most of these contributions are tagged "metadata" "socialnetworking" and "apml", and are rated high/have lots of views/backtrack links/tweets than it means that this author is an expert on these matter, within the confinement of the people who see/rate/tag the content (which can be a public of private network).

Have not yet published anything (actually, I only started this week), but this is what I hope to achieve: a kind of model or description of all different kinds of (user-generated) metadata and their relationship with quality, context, and expertise. APML like standards (or industry standards) can then be derived from it.

More info can be found in a proposal I wrote last year, in which I mention APML as well; http://www.scribd.com/doc/6370007/NWOMVI-Proposal-020908

I hope to see some fruitful conversations in the future on this matter and virtual-id issues, and am very happy to see some movement in this group.

Best regards,
Thieme Hennis
Delft University of Technology

[Renato Iannella]

On 17 Jul 2009, at 17:06, Paul Jones wrote:

I believe the key issue that needs to be resolved is as to how best to strike the balance between making the data easy to produce, and making the data useful to consume.

Simplicity is always key.  There is the possibility of capturing interests via vCard 4.0 CATEGORIES attribute:

See below....

Cheers Renato

NICTA

-------

I would like to add some extension to the 7.7.1 CATEGORIES property to make it relevant to capturing an individuals interests to supplement the category information. Specifically:

1 - Indicate a Preference/raking of values
2 - Indicate the source context
3 - Support URIs for linked data

1 - Preference
=========

Can we add the PERF parameter to the property definition. This will enable support to indicate strongly preferred categories over others.

Example:
  CATEGORIES:PREF=1;web science
  CATEGORIES:PREF=5;information management

2 - Source
=======

Can we add the PID parameter to the property definition. This will enable the linking between the category values and the source creator.

Example:
 CATEGORIES:PID=1.1;web science, privacy, social networks
 CATEGORIES:PID=2.2;soccer, field hockey
 ...
 CLIENTPIDMAP:1:urn:amazon.com:book:interests
 CLIENTPIDMAP:1:urn:getty.com:aat

3 - URIs
=====

Can we update the Value Type to also include URI values. This will enable the value to be directly linked to well known online vocabularies.

Example:
 CATEGORIES:Value=URI;urn:word.net:travel-agent

======

[Gérard Dupont]

I'm not sure but don't we forsee to connect to FOAF model too ? I mean, this is in my sense similar to vCard but in the "semantic" world.

--
Gérard Dupont
Information Processing Control and Cognition (IPCC) - EADS DS
http://weblab-project.org

Document & Learning team - LITIS Laboratory

[Paul Jones]

2009/7/20 Gérard Dupont <ger.d...@gmail.com>

I'm not sure but don't we forsee to connect to FOAF model too ? I mean, this is in my sense similar to vCard but in the "semantic" world.

This is indeed exactly what I was referring to when talking about needing to decide which path we were going to go down. I recall that some members of the group were intending to work on a proposal for an RDF-based variant of the spec - if anyone working on that is able to provide any details on where this got to, it would be useful from a direction perspective.

On Mon, Jul 20, 2009 at 03:05, Renato Iannella <rian...@gmail.com> wrote:

Simplicity is always key.  There is the possibility of capturing interests via vCard 4.0 CATEGORIES attribute:

Indeed, simplicity is definitely important. I would like to avoid overloading existing technologies where they don't necessarily fit though. I'm also mindful of at least keeping the current design of APML somewhat in mind too, and not departing from it to a completely different representation semantic simply for the sake of it.

Gérard, you were saying that you had some ideas about APML going forward?

Paul.

[Renato Iannella]

On 20 Jul 2009, at 17:31, Gérard Dupont wrote:

I'm not sure but don't we forsee to connect to FOAF model too ? I mean, this is in my sense similar to vCard but in the "semantic" world.

A lot of the key aspects of FOAF can be done in vCard 4 (and since the big players will support vCard in the future, then there is a good reason to do this...)

And we are also refining the cvCard/RDF specs at the moment (to keep the SW people happy ;-)

Cheers

Renato Iannella

http://renato.iannella.it

[gdupont]

> Gérard, you were saying that you had some ideas about APML going forward?

Well my major idea, except the semantic format of APML, is to extend
the context of APML. Let me give some details on how I get to this
before explaining what it could be.

- What was APML designed for ? Compiling a view of user interests
computed by a (web) application given some bunch of history which can
be seen as interaction with content.
- Why is it no so used ? It has been submitted some times ago now and
we do not see the real use of such data because the "compilation" is
hidden behind the application and so it is very difficult to compare
the score assigned on some content from application from the score
assigned to application B on the same content...
- How to overcome this issue ? First idea already proposed is to
expose the compilation strategy in the APML through some kind of
description of the mathematical model applied or the probability
function of the outcome. Obviously, application will be reluctant to
do so.

There comes my last idea (with a very nice introduction despite the
fact that it is not really that brilliant) : include in the APML some
generalized history of user interaction with content. This is what
google proposes through Web Search History which can be exported in
RSS. Having a common expression model on such data in in my sense very
valuable and enables any application to exploit it for it own use. One
of the major purpose will be of course to fuse information from
different application in a common way such as my web radio player, my
mp3 player...

More or less this could be a rationale to publish and share this "user
log" which are now under lot of attention due to privacy issue. In my
sense it will be valuable for any application provider that do user
logging to publish their data as it for instance to give user control
on the data they provide to the application (so again similar to
google web search history) and evnetually for export to other
application that can be enriched by such data.

Any comments ?