- From: Harry Halpin <hhalpin@ibiblio.org>
- Date: Thu, 24 Aug 2006 17:59:28 +0100
- To: public-grddl-wg <public-grddl-wg@w3.org>
Everyone, Since it's the duty of every member of the working group to make sure our publications are of the highest quality and readable, please do take the spare time read through Editor's Drafts as they are produced. However - all these comments are only comments, and it's up to the editor's discretion what to include or not. Here's my current review of the Use-Case document. Overall, great work - I'm very pleased we've got some simple use-cases as well as cutting-edge ones (wiki, Xforms) - in addition to outreach to user communities like e-learning and health care. In general, I'm going to add that the use-cases don't have to be fully coded, but if they are related to projects of yours or projects you're working on, you can add a link in the text/bibliography. So even if a use case isn't in the primer, if we get the code or a more detailed tutorial working for a use-case (or even an academic publication), the user can find more info. Second, we should remember to prune and re-order the use-case document a bit. I think ordering the use-cases from easiest to more challenging is probably the way to go. So, my revised order would be 1. scheduling 2. clinical data 3. buying a guitar 4. digital libraries/W3C, and then the wiki and Xforms cases in some order, with the "live bookmarks" case being dropped unless further developed, the structured query catalog use-case being dropped (since it's kinda close to buying a guitar), and the meeting extraction use case being put at the end until BenA gets back. I'll skip scheduling, and go to the Wiki use-case. It sounds great, yet what I would like to hear would be how a particular human teacher at the Univ. of Marcilly is going to use this semantic wiki and GRDDL to solve a concrete problem that previous XML/HTML technologies w/o RDF couldn't let them do. It seems that the problem that Franck faced earlier in the e-learning use case would be one way. Let's say Franck is teaching a course on a concrete topic like linear algebra or history and wants to dynamically generate his course materials. He wants to allow his students to comment and add to his slides and papers, as well as post new links to relevant subject materials, so he wants to use a Wiki as well. So, let's say he's teaching a course on French History. If various slides and papers are marked up with time-spans, he could dynamically put together a page on French history on the French Revolution. He could do this by specifying "France" and "Years 1789-1799". But maybe he wants to teach a class on revolutions in general, so he wants to include "USA" and "1775-1776" in his keywords. Or he wants to see what interesting things were happening in France during the American revoluion, so he could use "France" and "1775-1776". Or perhaps he wants to move up the class hierarchy and look at all "Europe" materials from "1775-1776". He might also want to use RDF to distinguish between his authoriative (perhaps uneditable) work and the comments and links his students add. Just an idea on a concrete example of how using RDF to dynamically restructure wiki data can produce better e-learning pages. I'm a bit less comfortable trying to provide more motivating padding for the XForms Atom example, because I've never made an XForms app, although I have used them and they're very cool. So, I'm a bit confused - it seems like what the use-case is that Voltaire (great name!) is using this XForms/Atom combo to automatically update their blog with the content of publishers with similar interest. So, let's say Voltaire has a blog about bird-watching. However, he's busy watching birds and doesn't have time to surf the net and find feeds of his friends as he looks for kites, a type of bird. The professional ornithologist Johan Bos, who recently spotted a red kite (Milvus milvus) far from their breeding ground in central Wales, and who recently posted a blog entry about it. However, (is this what RDF Forms in combo with the Intropsection document do - I'm not really seeing how this is possible...) when Johan updates his web-log about the red kite, it automatically gets posted to Voltaire's blog. Is this correct? Hmmm...why shouldn't Voltaire just subscribe to Johan's blog feed, using just normal XML? Is it the posting operation? A bit confused. As for the guitar use-case, I'll leave this in Brian Suda's hands - but it would be *great* if we had a real-live microformat hReview database to make this work, instead of having to imagine one. And then, as Danny Ayers said, we have to "pretend" they authorized the use of GRDDL on their data. With the W3C Use-case - I'm a bit unsure how normal people could see the value of this unless it's emphasized the RDF is turned *back* into HTML after rule-processing using XSLT. Otherwise, it appears that we've just made a bunch of RDF from our tech reports - and unless you're an RDF fanatic, you won't really care. It's the conversion back to HTML after customized processing that makes this use-case work for normal users, librarians, etc. For the use-case involving clinical data, again - I'd like to see some more concrete details. Not tech details in abstract, but how a human facing a concrete problem would use the technology to solve it. Let's say Kayode (which competes with the University of Marcilly for a cool name) wants to query the data. What does the RDF buy him that querying using Xquery doesn't? Is the same information (like diagnosis type) spread out through multiple places in XML files that use differing schemas, so that no single XPath expression can be guaranteed to get all the diagnosis types? Are their differing XML files? In which case, then mapping to a unified ontology (like the one you said was Helen Chen from Agfa was developing - could you add a reference to this ontology?) would make sense, and this mapping could be done via GRDDL. Then, he could query the clinical data. I'm also going to note this is the XML-version of the HTML guitar-buying use case. Hope this helps, cheers, harry -- -harry Harry Halpin, University of Edinburgh http://www.ibiblio.org/hhalpin 6B522426
Received on Thursday, 24 August 2006 16:59:44 UTC