Ivan’s private site

A bit of artificial mental jumps here but, well, this is how mind works…

I had some fun with Zitgist’s Semantic Web browser. Great to see these things coming up one after the other (Fred should probably add an entry to the SW tool list when he feels it is mature enough…). The templating mechanism on Zitgist is clearly powerful to make the output more enjoyable for humans (will I be able to add/use my own templates, b.t.w.? I wonder. I hope…).

Frederick used my publication list (in RDF) as an example in his blog on the browser which leads me to a related subject. In one of his emails (or was it IRC?) Fred referred to a possible project on bibliography ontology, which reminded me of an older blog I had on bibtex and RDF (and Bibsonomy) a while ago. Well, the publication list referred to and (nicely!) displayed in the Zitgist browser is actually a static dump of the RDF bibliography generated by Bibsonomy. And the good thing is that some of the grudges that I had in January are gone (authors and editors are displayed in correct order).

So what is Zotero, and what does it have to do with this? I discovered this tool about a month ago and I have been using it since. It is a Firefox extension to manage, organize, maintain collections of research resources, eg, article references. One can easily add notes, define cross-relationships, tags, add attachments, etc, to article references. The scholarly references can then be exported in various formats, including RDF. Yes, it is yet another RDF format for bibliography (see my note of January again…): at the moment the generated RDF is a mixture of dublin core and FOAF tags (plus some local tags). But, as far as I know, the Zotero developers want to work on this part of their system further (Zotero is still in Beta). I think that if some work on bibliographies is really picked up by the community the Zotero developers should be included somehow, too. Zotero might become a very powerful tool for data exchange in the the scholarly research community, would be good to have them on the (SW) board all the way! It would be nice if there was an easy connection and data exchange between Bibsonomy (which is great to maintain bibliography on line and share them with your peers) and Zotero (which I regard more as a personal management tool).

David Huynh sent me a mail on a cool feature he has just added to Exhibit 2.0: he’s put a shortcut to the Babel service. What this means is that one can refer to an RDF/XML, Excel, or a BibTex file instead of the JSON code, and Exhibit will convert it to RDF on the fly. So, to try it out, I put together a small example: the list of the W3C related talks of my buddies, ie, people whose name appear in my foaf file. And the remarkable thing is that this needed around 20-25 minutes (including chasing some stupid misspelling in the SPARQL query:-). And here is how it works:

First of all, the W3C Talks’ system is based on RDF: all data are stored in (public) RDF files and the interface, the query, etc, is just a front-end to the data. (The links to the relevant RDF files can be found at the bottom of the Talks’ page). I could therefore come up with a SPARQL query which looks more or less as follows (I omitted the namespace declarations from here):

CONSTRUCT {
    [] rdf:type talk:Talk;
       dc:date ?date;
       foaf:name ?name;
       talk:event ?event;
       vcard:Locality ?city;
       vcard:Country  ?country;
       dc:title ?title.
}
FROM <http://www.w3.org/2004/08/TalkFiles/2006/Talks.rdf>
FROM <http://www.w3.org/2004/08/TalkFiles/2006/WWW2006.rdf>
FROM <http://www.w3.org/2004/08/TalkFiles/2006/BeijingEvent.rdf>
FROM <http://www.w3.org/2004/08/TalkFiles/2007/Talks.rdf>
FROM <http://www.w3.org/2004/08/TalkFiles/2007/WWW2007.rdf>
FROM NAMED <http://www.ivan-herman.net/foaf.rdf>
WHERE {
    GRAPH <http://www.ivan-herman.net/foaf.rdf> {
        [] foaf:knows [ foaf:name ?name ].
    }
    ?talk dc:title ?title;
          dc:date ?date;
          talk:presenter ?person;
          talk:event [
              ical:description ?event;
              vcard:Locality   ?city;
              vcard:Country   ?country
          ].
    ?person contact:fullName ?name.
}

This query can be run on a SPARQL endpoint (I used SPARQLer for this one) via the SPARQL Protocol, using a suitable URI. The URI is actually pretty ugly, so I made a tiny url out of it, and I simply plugged this tiny uri as a data source for Exhibit in the HTML header. And, voilà!, the data displays. A bit of tweaks on the Exhibit lenses to make it a little bit nicer, define the facets, that sort of things, and put it on the Web. And the beauty is the typical mashup effect: if I add a new person to my foaf file or a new entry is added to the W3C Talks’ data, the display remains up-to-date… All data used here are pure RDF/XML files but, if the SPARQL endpoint understood, say, GRDDL on the fly, then I could have used my W3C Homepage as a datasource, too…

One issue that I find interesting/important here is the crucial role of the CONSTRUCT facility of SPARQL, which adapts the output of the Query to Exhibit. I have the impression that, when talking about SPARQL, we do not emphasize enough what a great feature this CONSTRUCT is. A SPARQL endpoint acts as an RDF transformation engine which is really powerful.