(by Frank van Harmelen)

Instead of producing my own write-up of last week’s conference, I’ll point to the two best blog entries on the conference from elsewhere:

A very good write-up of the general spirit and the highlights by Ivan Herman (W3C): http://ivanherman.wordpress.com/2008/10/31/iswc2008-karlsruhe/

A good identification that topics central to LarKC are rapidly taking central stage at the conference is by Pascal Hitzler, at  http://blog.semantic-web.at/2008/10/30/a-very-personal-bit-of-iswc08-trendspotting/ Don’t take everything he says seriously ;-), but I agree with his core observations.

My own observations about the conference were:

After the “breakthrough” feeling I had after Busan07 (where I saw really large datasets appearing for the first time, the birth of the BTC,etc), this conference felt like “steady progress”. There were very good scientific quality, the restrictive policy of the PC chairs paid off; I found (almost) no paper where I thought “why did this get accepted”.

The Semantic Web Challenge remains a highpoint of the conference (as in many previous years), and remains a very good thermometer of what is possible with the current state of the art. It was impressive to see that all submissions now use very large, realistic and often external datasources, and that all submissions come with credible user-interfaces. Of course we were delighted (and somewhat surprised that MaRVIN won 3rd price in the Billion Triple Challenge).

It was also very noteworthy how full the rooms were in the industry and in-use sessions. Very interesting to see that apparently the SemWeb crowd cares a lot about this (I’ve seen very differently in other academic communities…)

There were remarkably few papers on web-services, what to think about this…

The panel on “An OWL 2 far?” turned into a bit of a rerun of the familiar arguments concerning the design of OWL: how much expressivity is required/harmful, is OWL serving the right set of use-cases, should OWL be enriched (as in OWL2) or rather dumbed down (as I would suggest), etc. For many involved in the design of OWL it was a bit of a deja vu, although I spoke with some newer members in the community who found the explicit discussion on design trade-offs enlightening. 

Dealing with uncertainty is gaining its place on the agenda (and rightly so, of course). Unfortunately, most of this work is staunchly in the numerical camp (e.g probabilistic DL), our own work on Rough DL is still unique (or: lonely…) in adopting a qualitative approach to modelling uncertainty and fuzziness. 

Points of specific relevance for LarKC:

• It’s becoming very clear that the Semantic Web is the “Web of Data”, and not/no longer the “Web of ontologies”. Ontologies are (only) there to help with organising/querying/aligning/reasoning-about the data.
• It’s very good to see that those data are now becoming rapidly available. The Linked Open Data cloud is becoming increasingly frequently used. 
• As a result, scale is becoming a real issue (the success/growth of the Semantic Web is now becoming a problem, so that’s good). The workshop on scalability was apparently very good (unfortunately I didn’t attend it). 
• As one way towards scalability, approximation is attracting more and more attention.

All of the above 4 points are very good news for LarKC, and are making LarKC only more relevant than it already was. Of course, this also means that LarKC is less far “ahead of the crowd” than we were when we wrote the proposal (1.5 years ago, by now). On the good side, this means that LarKC will attract more attention and can have more impact (assuming we get it right, of course). 

Some noteworthy papers for LarKC:

• Identifying Potentially Important Concepts and Relations in an Ontology, Gang Wu, Juanzi Li, Ling Feng, and Kehong Wang. This is closely related to work that Enrico Motta is doing on identifying important concepts (aka  ”summarising an ontology”). I’ve also seen similar work by others. This is very relevant for LarKC’s IDENTIFY component
• Laconic and Precise Justifications in OWL, by  Horridge,  Parsia, Sattler. This is very relevant for LarKC’s SELECTion component
• MarK Greaves in his talk identified the “Web of Data” as the “2nd Semantic Web wave” (with the first wave being database integration). This 2nd wave is where currently all the technical action is (he’s right about that). He also pointed out that the business-cases for this 2nd wave are still very poorly understood. Notice that this 2nd wave exactly where LarKC could come in its own. 
•  the paper on “Anytime Query Answering in RDF through Evolutionary Algorithms” by our very own Eyal, Christoph and Stefan was well received as very innovative and promising (although we acknowledge that the promises are currently better than the results!)
• An Interface-based Ontology Modularization Framework for Knowledge Encapsulation, by 
  Faezeh Ensan and Weichang Du is relevant for LarKC’s SELECTion task

This entry was posted on Friday, November 7th, 2008 at 8:50 pm and is filed under Uncategorized. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.