Ivan's Blog

There is an official news item for this, no reason to repeat it here… but there are details that are really worth mentioning. Of course, the emphasis are mine.

The group developing GRDDL will provide a long awaited “bridge” between RDF and microformats (or, depending on where you come from, between microformats and RDF…). The nice thing about GRDDL is that it is simple. Of course somebody will have to write the transformations from a specific microformat to RDF, but GRDDL makes it possible to develop generic processors that “know” what to do, where to fetch the transformation, etc. The very little price to pay is to add an XTHML profile attribute and a pointer to the transformation itself; the rest should be automatic. Details are of course to be worked out but the group starts with an existing note already, worth looking at it.

The new Semantic Web Deployment Group includes several interesting work items; just two of those:

(1) RDFa (see also the separate page for further info) is very similar to microformats but provides a more scalable mechanism. Whereas microformats are a wonderfully simple and efficient way to add a specific set of terms to an XTHML page as, essentially, metadata, RDFa makes it possible to do something similar, but easily mix various vocabularies just like… well, RDF does in general. RDFa requires a bit more rigour but it provides a good alternative to microformats for more complex applications. And it will be possible to define a general GRDDL-like (or simply GRDDL) mechanism to extract the RDF automatically. (Actually, such tools already exist.)

B.t.w., RDFa and microformats+GRDDL can happily live side-by-side for different application classes, there is no need to present them as antagonistic technologies… (And no, it is not true, RDFa is not bound to XHTML2! Look at Mark Birbeck’s note [but see also the note below].)

(2) SKOS is something different: it provides a way to “bridge” (again…) to the more traditional library world, for example, to bring glossaries, thesauri, etc, to RDF. Eventually, one should be able to use SPARQL to federate several of those thesauri on the Web. SKOS does not have the same complexity as OWL, but for some applications this is exactly what one needs. I remember Alistair Miles, one of the drivers of SKOS, saying “OWL is the sledgehammer, SKOS is the nutcracker”. Although SKOS is not yet a Recommendation, there are already quite a number of people experimenting with it around the World. (Just one of the many examples: I met a guy in Australia last week who uses SKOS in Canberra to model medical data…)

Something which is not (yet) in the current Activity, though W3C got several comments on it, is the relationships between Relational Databases (or SQL, if you like) and RDF. How exactly would you map data from, say, mysql to RDF? How would you describe the mapping: is it some sort of a specialized ontology, or is it a set of (Horn) rules (or the combination of the two)? How would you name a record in a Database, ie, what is the URI to be used for RDF? These are all exciting issues and very important for the future of the Semantic Web. The reason it has not been included in the current round of W3C Groups yet is because… well, frankly, there are too many issues to solve before it is mature for standardization. But something is moving: a separate wiki page has been set up on W3C’s site to collect information; it is worth sharing information and experiences. Actually, W3C might organize a workshop on the issue sometimes late 2006 or early 2007. So: watch this space!

Benjamin Nowack pointed out on #swig that Mark’s note is entirely correct because some of the features he cites in his blog are, in fact, not compatible with XHTML1.1. And Benjamin is correct. What is important, though, is that the goal is to decouple RFDa (was: RDF/A) from the development of XTHML2.0 and, instead, define it an XTHML1.1 module.

This may be only marginally interesting for the Semantics Web, but who knows…

A colleague of mine, Karl Dubost, drew my attention on the presentation of Hans Rosling. Hans Rosling comes from the Karolinska Institute in Stockholm and made a very nice presentation on the evolution of health and other statistical data in various countries around the globe. The data itself is very interesting, but what may be even more interesting is the methods he has to show those. With some colleagues at the Gapminder foundation he developed great visualization tools to make boring statistical data lively. Worth having a look! Statistics may not be as boring as one might think…

Beyond the interest in information visualization, why is this of any interest for the Semantic Web? Well, at the end of his talk he refers to the problem of data integration, of the difficulties but, at the same time, the importance and the interest of integrating data coming from different (statistical) sources. And that is the core of what Semantic Web is all about, isn’s it? If somebody has a contact to the Karolinska, it might be interesting to look at possible connections…