See also: IRC log
<TomB> Previous: http://www.w3.org/2008/05/27-swd-minutes.html
<TomB> Scribe: Elisa
Tom: Considering pushing this week's agenda to next week due to low turnout
<aliman> http://www.w3.org/2006/07/SWD/wiki/SKOS/Reference/Planning
Alistair: All of the content is there; We need roughly 1/2 day to make a
few minor changes, add schema files, and that's it. There are a couple of issues in the schema files
that Sean sent, inconsistencies, that may have been due to transforms or tools.
... When I was using tripler before, I got some weird behavior that I wasn't able to pin down. The problems were minor, so I
just need another 1/2 day to get the files together and send them to Ralph.
Tom: Let's do that on the list then. We did decide last week that the changes had been incorporated ... whether we need to wait until next week or can do that earlier, let's wait until the files are posted and then take a decision online.
<benadida> Hi all, unfortunately I can only join via IRC today.
Ralph: Do we have a resolution to publish?
Tom: Yes, if you look at the 5-27 minutes, at the end of the SKOS section, the discussion was to publish the draft without formal review.
Ralph: Yes
Tom: I would propose that that also means that we could take the decision online and wouldn't need to wait for the next telecon.
Ralph: I concur with that.
Tom: We could close issue 83, but I think the call is too small, so I don't see any other items on which we could make significant progress this week. I suggest that we resolve to use this agenda as the starting point for next weeks call, unless anyone has issues on which we think we could make progress with such a small call.
Ralph: What I'm wondering - Ed, do you know where Jon is making updates where stuff is fixed?
Ed: He seems to be pushing them through CVS.
<JonP> I am
Ralph: I'm not seeing all of the changes.
<Ralph> Jon, is there an updated .html document?
Ralph: ... because some of the changes you had suggested consistently used the same domain name ...
Ed: Alot of the things I was pointing out had to do with the files
... I did have a question about ... he was using mydomain in one of the examples.
<Ralph> Jon, e.g. 'yourdomain' vs 'example.com'
Ralph: That was one of the questions I wanted to raise.
<JonP> There haven't been any updates to the html since we posted the last draft.
<Ralph> (W3C pubrules prefers either w3.org or example.com)
<JonP> ...trying to get on the call
Tom: I wanted to talk about some of the things that might continue after the working group, as a question to Ralph, what are the opportunities, resources available to ... as far as use of wikis is concerned, support for mailing lists, what are the options?
Ralph: If the working group wants to ask for or sponsor a SKOS interest list? I don't have a feeling either way whether that's valuable or not. We can certainly host those lists and the wiki as well.
If we can, snapshot the state of the working group, not necessarily freeze it, but make the point the working group reached clear for a couple of years from now... I would prefer generally that we establish a separate set of SKOS wiki pages
Alistair: That would be nice.
Tom: Maybe the current wiki pages could be frozen.
Ralph: The esw wiki will at some point be migrated over to ... Ralph, so the current difference in wiki platforms will eventually go away ... in a single digit number of months.
Tom: back to recipes ...
Ralph: I did write an email shortly before the telecon -- it's useful for us to be careful about consistency. That's why I asked if the updates you're making are visible somewhere ... if the served example files are being updated somewhere.
Jon: It's enough of a pain to update everything that I'm staying with the last date where I can. I think I only changed one small thing in that document and that was a date. I haven't changed the example.org changes until you can do a review of the document.
Ralph: I guess I'd like, if you would create a 2008-06-xx.html document, the dates can be inconsistent.
Jon: I haven't actually changed anything in the document since the draft that I put up, almost all of the changes I can think of are in.
Ralph: So the only thing you can think of is the yourhost ...
Jon: What's the process at this point? Does this need to go thru any additional review?
Ralph: If Ed and I say that this is ok, then the working group should be ok with it.
Jon: if I make the change and put up the document, it's ok to reference the 2008-04 examples, without having to change all of the dates?
Ralph: Yes.
<Ralph> (for an updated editor's draft)
Jon: I just have to make the one change to the document and repost it.
<Zakim> aliman, you wanted to ask about URI dereference behaviour for SKOS concepts, schemes etc.
Alistair: I was just looking ahead at the SKOS issues, there is one on the URI dereference behavior, and I was wondering if anyone had any ideas about how to resolve this relatively painlessly. Perhaps we can resolve this to a recipes example.
Ralph: I definitely thought we would go in the direction of working to a recipe.
<Ralph> issue-86; SkosURIDereferenceBehaviour
Alistair: 6 is the extended configuration for dereferencing to a concept.
Ralph: I don't necessarily think we should resolve to recipe 6 in particular ...
Ed: You can imagine someone opting for a simpler one.
<Ralph> LCSH
Alistair: What do you do for LCSH?
... They have a way of working around doing 303s but still being able to dereference each URI.
<aliman> http://lcsh.info/123456#concept
Ralph: How much trouble would I be causing to ask for a write-up for this ...
Ed: It was originally done using one of the recipes, but this simplified the server side code. I would be happy to write it up, but it would probably need some time to put it together.
<aliman> ... where http://lcsh.info/123456 is directly negotiable
Alistair: From what I understand, I think its ok, but you do have to be a little bit careful, as there is a primary URI, but the risk is that if you serve html content, and if that had an anchor in it for your concept ... you could end up with a conflict ...
Ed: I've been trying to make sure that the html references do have an anchor with concept, so maybe this is an issue.
<edsu> http://lcsh.info/sh85118553#concept
Ralph: When I do an http get on an URI which is a subject heading, it's something that has (see example) ... if I do an http get on that, the server may not respond with a 200, as the subject heading is not a document.
Alistair: I'm not sure that's true, because you can't do an http get on that directly, you need to do that on the primary URI.
Ralph: This is exactly the point you made about the difference between the html and RDF version, and in this particular design, the server can't say 303 for a concept, because it doesn't know you asked for the concept. It creates additional confusion if the html document has anchors called concept and the RDF has ids called concept. This means that the TAG will think that the LCSH approach is broken.
Ed: I definitely understand the issue of having the hash URI that has two different representations, I guess I was encouraging Jon to promulgate this error ... :).
Alistair: If there was not an anchor in the html, I thought this could be considered a new recipe, as long as there wasn't a direct contradiction to web architecture.
Ralph: That's the difference between recipes 1 and 3 ... the crucial difference is that recipe 1 doesn't have any html and recipe 3 introduces html and a 303.
Ed: That's originally what I did.
Ralph: I'm afraid that doesn't work ... appreciating Alistair's attempt to massage it.
<TomB> http://www.openarchives.org/ore/0.9/primer.html#HTTP
Alistair: The argument was really fine, all of the recipes are more cautious with respect to 303s.
<TomB> http://www.openarchives.org/ore/0.9/primer.html#Aggregation
Ralph: It creates dragons for revisions to the vocabulary - you have to be sure not to introduce any anchors.
<Ralph> Cool URIs for the Semantic Web
Ed: In the Cool URIs document, there was feedback from the TAG ... there was something in that document that shifted in its tone at some point.
Alistair: The feedback I had from TimBL had to do with generic resources, from what I understand, Tim was saying do a 303, and then content negotiate from that point. If you have a generic resource then you can bookmark it. I thought back through what we have done ... you can't go back through and retrofit that to a document, you would have to publish a new document.
<edsu> http://www.dfki.uni-kl.de/~sauermann/2006/11/cooluris/
Ed: This looks like it was the old document, but if you look at figure 4, it doesn't look like what we're doing now. Look at figure 4, so it's in section 4.1, the hash URI section.
<Ralph> Ralph: the dfki version is old
Alistair: That's interesting. Then you would be able to fix it by making sure that the content location header is correct.
<aliman> The Content-Location header should be set to indicate if the hash URI refers to a part of the HTML document or RDF document.
Ed: I'm open to changing it if we have a recommendation. I'm willing to change it appropriately, and that's the reason I put this experimental version out. Maintenance over time is definitely a big issue.
Alistair: There's quite a bit of difference between the Cool URIs document and the recipes we're putting out.
Ed: I think that was introduced in the stuff that Diego noticed.
Alistair: We've only got a recipe for the first part of 4.1 and 4.3.
Ed: But you started this a long time before this came out.
Ralph: We might need to say something about this other document with additional cases.
Alistair: But the recipes document would get considerably more complicated
Ralph: We should at least acknowledge the one more useful pattern.
Ed: The recipes document does point to Cool URIs. This has come up before, the potential for RDFa to deliver vocabularies.
<TomB> Cool URIs or "simple implementation": http://www.openarchives.org/ore/0.9/primer.html#HTTP
If you did use RDFa to do that, you probably wouldn't be able to deliver an RDF/XML graph at the same URI.
<Zakim> TomB, you wanted to mention OAI ORE example
Ralph: It will be awhile before we can recommend best practices around this.
Tom: I just wanted to quickly point out that if you click on the link to the OAI ORE example, it looks like they have two ways of implementing, using the Cool URIs or using a resource aggregation, but that's intended to resolve to the information resource, not to use the 303 redirect. I mention this because OAI ORE on the one hand, LCSH on the other ... these are two implementations people will look at for best practices. It would be good to make sure that the right precedences are set for these and documented.
Tom: Could you take an action, Ed, to write a note to the list? It's worth trying to nail it down and try to get the reasoning nailed down.
<scribe> ACTION: Ed to write up the rationale behind the dereferencing behavior for LCSH [recorded in http://www.w3.org/2008/06/03-swd-minutes.html#action01]
Ed: and Alistair can respond
Tom: I think that's going to be one of the implementation examples for SKOS, so we need to make sure everything is in place for doing the right thing for the recipes.
<edsu> aliman++
<edsu> # for bringing that issue up
Tom: Meeting Adjourned
Change log:
$Log: 03-swd-minutes.html,v $ Revision 1.3 2008/06/10 12:34:34 swick Copy new and pending actions from 27-May telecon Revision 1.2 2008/06/10 12:31:08 swick From Elisa Sun, 08 Jun 2008 11:42:50 -0700