Home > bossam, news > A new Bossam on Feb. 5

A new Bossam on Feb. 5

February 2, 2007 Leave a comment Go to comments

A new Bossam release is planned on Feb. 5, 2007.

The new release will include better OWL DL reasoning supports:

  • Enhanced coverage on classification based on boolean compositions of classes
  • Enhanced coverage on classification based on nominals (owl:oneOf)

Also, a new rule construct is introduced: rule-generating rules (RGR). With RGR, it’s possible to assert a rule at the consequent as shown below.


rule r is
if causality(?cause, ?result)
then
assert if ?cause then ?result;

RGR will help representing more complex knowledge and optimizing knowledge structure!

Also on performance-side, the RETE core of Bossam is well optimized and it gives a lot better performance than before!

Advertisements
Categories: bossam, news
  1. February 19, 2007 at 2:54 am

    I’m curious, does Bossom use the same approach as Jena2, Pychinko and Fuxi? I’ve studied those implementation and none of them are really RETE. They’ve mis-applied Doorenbos RETE-UL to RDF triples. Using a direct EAV (entity attribute value) mapping approach produces horrible performance in Jena2, Pychinko and Fuxi.

    peter lin

  2. February 20, 2007 at 1:43 am

    Well actually, I don’t know much about RETE-UL, so I cannot comment whether Bossam uses the same approach as the tools you mentioned. I’ll have to read more of your blog articles. At least, in Bossam, the RETE core has no direct connection to RDF triples. I’m working on optimizing Bossam’s performance on RDF/OWL reasoning, and I’m getting promising results, fortunately! Cheers!

  3. February 21, 2007 at 9:39 am

    That’s good to hear. It means you didn’t go down the same route as the other RDF/OWL projects. Basically, the other projects mapped subject-predicate-object directly to entity-attribute-value. Since RDF primarily deal with entities, it fits poorly with a 3 tuple (aka triple) memory representation.

    The better approach is to use a deffact like n-tuple approach, which is closer to Dr. Forgy’s original thesis and subsequent papers. I have several entries that explore the impact of using a triple design for RETE and show that’s only time it’s a good choice is when the facts naturally fit into a triple form. Although RDF expresses relationship in subject-predicate-object form, it is primarily reasoning over n-tuple entities. hope that helps

    peter

  4. February 22, 2007 at 12:12 am

    I understand. You’re right. In Bossam, there’s no distinction between RDF triples and general n-ary facts. I guess we’re in the same league. 😉

  5. April 15, 2009 at 5:59 am

    After reading through the article, I feel that I really need more information on the topic. Could you suggest some resources ?

  6. July 11, 2010 at 7:25 am

    If you love watching NFL, then you perhaps also proclivity to hazard on the match, and especially on your favorite teams?

  7. August 28, 2010 at 2:59 pm

    пердуперденция коньки

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: