SmILE Semantic Information Systems and Language Engineering Group

In the first week of September (from 3rd till 7th), I attended the Reasoning Web 2007 summer school in Dresden. It was organized already for the third time by the REWERSE EU NoE; the previous editions were held on Malta and in Lisbon. The main objective of the series is “to provide a coherent introduction into Semantic Web methods and issues with a particular focus on reasoning” (citing a related OntoWorld wiki entry).

The program was generally good and very “meaty”. The presenters were usually really experienced and respected experts in the related fields. The whole first day was pure theory, giving a comprehensive overview of foundations of rule-based reasoning with some special focus on the techniques applied in the Semantic Web. I found the first day presentations really cool and well organized, though sometimes kind of demanding. It went from the basics of first order logic and logical programming to the model theory and declarative semantics of logical programs. The lesson about declarative semantics of various rule-based paradigms was especially tough - it basically compressed a content of one semester into two hours, so it was very difficult to follow all the presented stuff even if you were consulting more detailed descriptions in the lecture notes. Then there were two another lessons about operational semantics of the rule-based systems and about related computational complexity issues.

The second day started with description logics lecture, which was quite fun - it described only the basic ALC DL, so it was relatively easy to follow. Moreover, the talk went from simple things to more complex ones in very incremental, clear and illustrative way, providing running examples wherever needed - very enjoyable. The rest of the day was about reasoning in semantic wikis - not much new when compared to the presentation Sebastian Schaffert had in DERI couple of months ago.

The third day was rather short - in the morning there was a lecture about policy representation in the web using rules. A little more practical than this first day and half of the second day - it presented several possibilities and formalisms of policy representation (mainly based on OWL DL descriptions or rules) and then provided some insight on how these policies can be automatically executed and/or negotiated when more than one policy is to be followed (e.g. in case when a person is requesting an access to a resource, the resource provider has a policy to get credit card details first and the requesting person has another policy for providing his or her credit card details). Interesting presentation, but maybe it would be even more interesting if a hands-on session was a part of it. After the lecture, we went to see things like the largest green diamond in the world, the smallest carvings on a cherry stone and many other useless shining things made from gold, pearls, diamonds, crystal and poor elephants teeth - everything in a museum treasury exhibition called Green Vault. The social event continued with a dinner in a castle overseeing the river Elbe - nice end of the day it was.

Day four started with another more practically oriented session about reactive rules on the web. Basically, two formalisms were introduced and compared - ECA (event-condition-action) and productive rules, as being implemented by ILOG in their BRM (and maybe also other) systems. Everything can be summed up in few words - the former is about rules that execute an action when an event occurs and a condition holds, all this happening on the web. The latter is about dynamic execution of a specific Java code when a condition holds (thus preventing undesired hardwiring of the dynamic and user-configurable rule execution process). The presentation was rather shallow and dealt with really obvious and trivial things. Whenever there was something interesting (e.g. resolution of conflicting rules in the ECA approach), it was basically put aside. This “presentation of obvious” style approach continued even during the afternoon for couple of industry presentations. Nothing really exciting and novel there - for instance, I really was not interested in hearing about the advantages of semantics-enabled search over the traditional full-text. I visited the Zwinger gallery (see below) instead of staying for the rest of the day.

The last day was opened by a presentation of the W3C RIF (Rule Interchange Format) working group. Really interesting snapshot of the current topics being resolved and discussed in the heated discussion of this large group with many different perspectives represented (either academia or industrial, either DL or logical programming based, etc.). Then there was another industrial and PhD session, but I already had to leave, so I could not attend it (according to the proceedings, it was most likely nothing really crucial).

The overall impression from the summer school kind of reinforced my opinion about the leading Semantic Web reasoning paradigms (i.e. the DL-based and the rule-based in the logical programming sense). The theory is brilliant and the mathematics beyond them beautiful and challenging indeed. However, the practical Semantic Web applications and real world adoption of the results in this field of research barely go under the surface, if adopting any more sophisticated stuff at all. Sure there exist cool applications of the full-fledged DL or resolution-based inference, but these either do not scale, or are expensive to be maintained (since the expert involvement is needed), or both. Therefore the industry and/or general public seem to adopt mainly rather obvious and simple (thus not that powerful) stuff coming out from the basic research related to the “real” inference. So perhaps the most important subjective and selfish result of my attendance at the summer school was further justification of the motivations for my PhD research, which could start to (at least partially) fill in this “hole in the market”, supposing it will be successful, of course.

To get your own opinion and perhaps learn something more about reasoning on the web, you can browse the presentation slides from the summer school in the REASE catalog. However, it seems that only Axel’s rule interchange presentation and the foundations of rule-based query answering block presentations have been made available to date. Hopefully more are still to be uploaded there by the summer school organizers or the authors themselves. The (Springer LNCS) summer school proceedings are in general very high-profile and provide valuable additional and background information concerning the presented topics. I have the printed lecture notes on my desk, so feel free to come along and borrow it. You can also access the online version at SpringerLink.

Concerning the non-science aspects of the trip, it was very enjoyable, too. Dresden city center is really nice - even though I do not particularly like the baroque style, I must admit that there definitely are some pieces worth seeing (or even staring at them;)) there. To mention few such things, the central castle is really impressive and it has almost unbelievably many faces depending on the weather, time of the day and your observation position. There is also this Zwinger gallery - I thought I would be mainly bored when going there to see the old masters, since I’m rarely impressed by the fine art before, let’s say, van Gogh. And it was true for some repeating series of relatively mediocre Italians or “local guys” (but to make it clear, I do not consider for example Durer as a this local guy, even though he was German:)). However, some pieces from Rembrandt and Rubens balanced the whole boredom perfectly. By the way, you can see Zwinger in the Second Life game - it is the first top-rank gallery present in the game, at least this is what I saw printed on a poster there (in the real, first-life Zwinger I mean). Yeah, and besides these cultural aspects, the German beer is almost as good as Czech. And there are many nice restaurants in the city, where you can get really tasty food (either local or international) to get your beer a company - I tested three of them more or less by random, one with German specialties, one Italian, one Thai, all were just perfect (and the prices too, at least when compared to Ireland:)).