Thinking Clearly .

This time on Logical Commonsense. Some good stuff from Davis as well as from the Common Sense Problesm page.

Now crack that egg!

The schedule for OWL Experiences and Directions is now up, and it looks good. Some of the papers are really very interesting, and all of them I think will contribute to good discussion.

Clark and Parsia (us!) are sponsoring it, though we’ve not given our new logo for the OWL ED website yet. I blame Kendall!

The most important session, in my mind, is toward the end of the first day, “Ongoing and Future Standardization Efforts”. Here is where we lay the groundwork of what we are going to do over the next year. For me, I hope in the next year we can reach broad, implemented consensus on SPARQL/DL and DL Safeish rules. Oh yeah, maybe get a new round of standardization going.

OWL 1.1 is proceeding apace. FaCT++ and Protege-OWL support it. Pellet and (prelim) Swoop support should be ready by OWL ED (go Evren go!). I’ve not heard from the Racer guys, but they had a bit of a head start.

In a very minorly unusual move, OWLED is rather actively soliciting demos and posters right up to the demo/poster session. In fact, if you get a one page description of your demo/poster in by Oct 20th, you still have a shot at the proceedings:

The workshop will have a session for software demos and “posters” (formal or informal) describing work in progress. All participants, even those without a paper in the program, are invited and encouraged to demo their tools and/or present a poster. Interested parties may send the organizers a one page description of the demo/poster before October 20th. Such a description is not required, but descriptions submitted prior to October 20th will be considered for inclusion in the proceedings.

So, come out!

I’m planning to play around a bit with natural language processing for knowledge acquisition. (Two areas of interest to me: One is helping me, a fairly sophisticated reader, analyze and manage the web pages I read, esp. for the purpose of producing reports, aka blog posts, about a topic. My playing with Diigo is in this vein. The other is a knowledge representation based or enhanced Wikipedia. These two are, obviously, not unrelated.)

My standard first move would be to get a good textbook and start working with Prolog since 1) Prolog has a rich history of use in NLP, 2) I like Prolog, and 3) I really like my favorite Prolog, SWI-Prolog, which has a pretty nice bundled set of libraries and tools. But I just spied NLTK which seems to be an excellent, pedagogically oriented toolkit. I was quite attracted by the tutorial for the older version, but it isn’t clear if it will be updated. (Alas, the NLTK-Lite tutorial is just a set of slides. These are welcome, of course! And I shall definitely read through the older tutorial as well.)

Cypher also seems neat. At least, and as usual, in principle.