First of all why the delay? Well, I typed this blog at the airport of Busan and stored it on my machine. So far so good. But then I arrived to Beijing (where I am now) and I wanted to post them on my blog… and I failed. It turns out that all blogs of the type
XXX.wordpress.com have, ehem, difficulties to be reached from China. Why? No idea. It took me several days to find a way to access my blog again and post the original text. It is a few days old; hopefully it is still ok…
The paper of Saggion et al. reported on an EU project (called MUSING) which applies Semantic Web techniques to business intelligence. Business intelligence is one of the possible “verticals”, ie, application areas of the Semantic Web that one hears about, but I have not seen that many specific examples yet. The reason why this may be an area of interest is obvious: it is really about gathering, integrating and analyzing very different datasets (public or otherwise). So it was good to see something more specific. These type of projects might be a good application examples for the Linking Open Data project results, too: some of the data this MUSING project relies on, for example, is already available there (although they did not realize it until now:-(
YARS2 has already found its way in the news around Semantic Web as one of the high-end triple store facilities. But it was good to hear about the technical details from Andreas Harth. A cluster of servers offering a distributed storage of many many millions of triples… Cool!
It was interesting that two new Semantic Web search and indexing engines were published sometimes last week more or less at the same time, namely Sindice and Falcons. And so happens that, on the one hand, Giovanni Tumerello and his team made a presentation on Sindice at ISWC2007 and, on the other hand, I have just arrived to Beijing for a Chinese Semantic Web Symposium where the development team of Falcons (from the Southeast University in Naijing) will also make a presentation. Nice coincidence. Staying with Sindice for now: it is really a cool tool. One aspect I liked is the extra functionality they have put in to retrieve resources via inverse functional properties. I am curious to see when extensions to Sindice, for example, will appear for Semantic Web browsers (the authors claim that is is very easy to do…)
I was interested by the paper of F. Abel et al., adding access control to an RDF query. What happens is as follows: (1) the access control policy is described in what looks like a simple rules language; (2) when an RDF query arrives, each triple pattern is analysed by a rule engine and the result is the extension/modification of the original RDF query (eg, filters or extra triple patterns are added). This modified query is then processed by a usual query engine (at the moment they did not implement it for SPARQL, but they plan to do it soon). This looks like a neat idea and a good use case for SPARQL (access control is certainly one of the very important issues that comes to the fore these day). This is all the more interesting because, now that SPARQL is a Proposed Recommendation (yey!), it will be interesting to collect use cases to see if extensions to SPARQL will become necessary or not (probably yes).
There was an interesting BOF on the third day (we should be grateful to Paola Di Maio who has pushed for it) on usability and Semantic Web. There were some discussions as for what this exactly means: for some, this not only includes the (human) usability issues of Semantic Web applications, but also the programmers’ usability of various SW API-s. (Although, personally, I believe these two issues are better kept separate.) It was decided to continue discussions on a separate mailing list and to try to build a more visible and active community around the issue. The (hitherto pretty much dormant) SW User Interaction mailing list, ran part of the W3C Semantic Web IG, will be used for that purpose. If you are interested, sign up!
The third day also included a panel with the catchy title “Prodigy or Sociopath: the Adolescent Semantic Web”. I very much liked the format of the panel. I am often bothered by panels where each panelist makes a short (well, sometimes not-so-short) presentation first and then they expect the audience to participate after 40-45 minutes into the panel’s time. That model often fails in practice. In this case the organizers formulated 5 statements (like for example “The SW mainly requires light ontologies”), each question was elaborated upon by one panelist, and the floor was open for discussion after each such statement. And this model worked well although, as far as I could see, nothing very controversial was said after all during the discussions. Maybe the question “We need social scientists to join the SW conferences” was the one generating most of the comments. (Although I think everybody agreed that some sort of stronger cooperation with social scientist is really essential. As far as I am concerned, this is the most interesting aspect of the Web Science Initiative, too.)
Last but not least: there was a great keynote by Chris Welty. The type of keynote that forces some sort of humility and self-analysis on what should be and shouldn’t be done and how; to realize that, well, everyone of us gets it wrong, maybe more often than not, and that should be all right. For me, the most important message was his “Better not Perfect” statement: we should recognize the value when something is really “better” and accept it, use it, disclose it, even if we know it is not “perfect”. Scientists (I must also say, mainly European scientists) have often the tendency to try to cover, eg, all the edge cases thoroughly, strive for aesthetics, etc, thereby often loosing both simplicity and speed of deployment and usage…
It was a good conference. Some areas were missing, although this may not be because it was not tried by the organizers but simply because it did not succeed (yet). For example, there was only one paper on relationships to (traditional) databases (the paper of W. Hu et al) although this area is really very important (check out the presentations of the W3C Workshop on the subject and consider the fact that, according to some statistics, almost 80% of the current Web content comes from databases…). Security, privacy, and related issues were also more or less missing, with the exception of the access control paper that I already mentioned (although there was a separate workshop on the subject, but there were no real presence at the core conference). I would have loved to see more application papers, too, but maybe that is asking too much from a mostly scientific conference. There were also less “buzz” than at ISWC2006 on the corridors, coffee breaks, etc; this may have been the result of both the setting (this conference was a bit lost in a major conference centre) and the predominance of Asian attendance who are known to be less outgoing than their European, Australian, or US colleages. But, again, it was a good conference and a good week!