The Sun BabelFish Blog

Logic as the science of Garbage Collection

Everybody knows that logic does not bring anything new to the table that we did not allready know. It just reveals the consequences inherent in our thougts. By logical transformations we can never get to know what the world is like. For that we need our senses: we need to look at the world. So what is logic for?

Well it occured to me recently just how much similarity there is between garbage collection and logic. Consider that you communicate with some dude over the internet but that the only thing you know about his is that his email address is hjs@bblfish.net and that he knows you. This can be written in N3 as

 _:p1 foaf:mbox ≤mailto:hjs@bblfish.net≥;
      foaf:knows :me .

Let us imagine that you later get to know some dude at a conference and on his slide he puts up a url to his home page. You immediately add this information to your database like this

_:p2 foaf:homePage ≤http://bblfish.net≥ .
 :me foaf:knows _:p2.

So now you have 4 statements in your database. On looking at the home page you notice that it contains a mailbox. So you add to your database

_:p2 foaf:mbox ≤mailto:hjs@bblfish.net≥ .

Now you have 5 statements in your database. And here logic kicks in. As foaf:knows is an owl:InverseFuntionalProperty (ie: for the same object there can only be one subject), you can deduce from

 _:p1 foaf:mbox ≤mailto:hjs@bblfish.net≥ .
 _:p2 foaf:mbox ≤mailto:hjs@bblfish.net≥ .

_:p1 owl:sameAs _:p2 .

Ie. you can merge (smush is the technical term) _:p1 and _:p2, and end up with the graph

 :me foaf:knows [ foaf:mbox ≤mailto:hjs@bblfish.net≥;
                  foaf:homePage ≤http://bblfish.net≥ ; 
                  foaf:knows :me ] .

Which says that I know someone with mailbox hjs@bblfish.net and home page http://bblfish.net and this person knows me. Which is 4 statements long instead of 5. Now if there had been more duplicate content on each of those nodes, then the reduction could have been a lot larger.

So what is an inferencing database? It's a database that can garbage collect redundant data.

[just a thought. see comments]

Posted at 03:27PM Jun 30, 2006 [permalink/trackback] by Henry Story in SemWeb  |  Comments[2]

James Gosling at Java Days in Paris

James Gosling came over to Java Days in the beautiful town of Versailles, just outside Paris. He gave his presentation on the state of Java, and of course answered the numerous questions from the panel and from the audience.

It is always nice to see James Gosling speak. He manages to answer some of the most difficult questions thrown at him. Of course someone always wants to know what is next for Java, and James's implacable reply is: "that's up to you, the community. Join the JCP." When he speaks one is reminded at just how far Java has come: It is used in billions of devices, is faster than C, and close to fortran, is used for business processes as well as for AI development, ... what a success!

James took the RER (the fast underground) back to his hotel in central Paris, and so I had time to chat with him about work in the Semantic Web. The really refreshing thing about James is his positive attitude combined with the breadth of his knowledge. For example he has no trouble mentioning the word AI (I believe I even heard him mention that word three times at the Java One keynote this year) . It's refreshing. Especially as he is able to bring some very interesting historical insights to the discussion. We talked for example about triple databases and quads (N3 graphs), and he pointed out how statements about statements were very useful in sensor devices, such as japanese rice cookers, that need to keep track of the measurements made by a number of sensors each of which with only a partial view of the world, and how fuzzy logic is used to help merge these different points of view to help make the perfect rice dish.

Posted at 02:42PM Jun 30, 2006 [permalink/trackback] by Henry Story in travel  | 

Java Days in Paris: Open Source on the front stage

Open Source Software (OSS) took the front stage at Java Day in Paris, yesterday. Five active Open Sourcereres answered questions as to

• what their involvement with OSS was ?
• why they had embarked on OSS projects ?
• how it had helped them ?
• what the business rationale for going Open Source were?

This was followed by questions from the room. Now the five panelists, all members of the Paris based OSSgtP group, were not just minor contributors. We had

• Marc-Antoine Guarrigue: works on Jcaptcha a library to help tell humans and computers apart, which if very useful for online service.
• Ludovic Dubost: creator of XWiki, a powerful wiki server that uses Groovy as a scripting language.
• Guillaume Laforge: the lead on the powerfully Groovy scripting language
• Vincent Massol: author of books on junit and maven and contributor to maven, cactus and cargo projects
• Emmanuel Bernard Contributor to the Hibernate Project (and helping me out with so(m)mer).

The idea here was not to come up with a general theory such as the excellent article by Simon Phipps on freedom to leave but rather to ask some practical questions to some people in the field, to help others feel more comfortable about the OSS notion, by having real people do the talking.

So to take Ludovic Dubost for example, creator of Xwiki. Ludo pointed out that there was no way he could have built Xwiki without the OSS movement. There just would not have been enough tools around, and if there had been they would have cost too much money for a product such as Wiki to find funding for. What banker would have put money on a wiki tool? Just imagine trying to explain to them what wikis are for: "Dear Banker, we want to write software so people can openly and freely share information"... Xwiki uses every other technology of the panelists: It uses Jcaptcha to test for robots, Groovy for the scripting language , Hibernate for database access, and of course everyone uses unit tests, and build tools such as Maven or cargo. Ludovic also started his own company to provide wiki hosting and gets paid for support and guarantees, as pretty much everyone on the panel. But one can find one's interests in OSS in much simpler ways. Just participating is good. It helps meet other people, is an excellent way to get exposure to good ideas and good people, and often helps find a job. Well that's how I ended up working at Sun :-) And coincidentally also how Bill Joy got his job.

Posted at 11:34AM Jun 30, 2006 [permalink/trackback] by Henry Story in travel  |  Comments[1]