Semantic Search - Faceted Search and Web 3.0 for Drupal

212 queries to load the front page of AHIRC. That'll do it.
This happens because I am trying to emulate the "COUNT" mixed with "GROUP BY" you'd find in SQL. One company has it:
http://docs.openlinksw.com/virtuoso/rdfsparqlaggregate.html
And looks like others are anxious:
http://simile.mit.edu/mail/ReadMsg?listId=14&msgId=16326
But it's not an easy problem in a broader context:
http://www.ldodds.com/wordtin/Wiki.jsp?page=UniqueNamesAssumption

Development is coming along, and there are a number of possibilities that Sesame opens up.

When you can run complex queries on Drupal data, maybe that will convince some skeptics of the value. Take for example:
http://www.openrdf.org/contrib.jsp
like
http://www.ontotext.com/bor/ (rules engine) and http://www.ontotext.com/owlim/ (OWL store and engine)

I've proposed to present the module at DrupalCon 2007 in Barcelona. I've got quite a bit of work to do before the official "dev" release (beta) - but solving the speed issues will help tremendously.

see http://barcelona2007.drupalcon.org/node/633

The holdup was a 20 second delay per request from the a bug in the Zend Framework (not officially described as a bug yet).
Everything seems to be fast now!

I still need to connect the adapter which will send the Drupal data to Sesame.

I've created a patch to the Zend Framework here (one line).
The patch is here http://semanticsearch.org/patch-zend-framework-socket-php-http-client
I documented the process here
http://www.openrdf.org/forum/mvnforum/viewthread?thread=1417

The client is nearing completion. I'm working on create, update, and Delete. I'll post a version of the file on this site when done
Thanks for the help from Sesame's project leader in the following forum thread.

http://www.openrdf.org/forum/mvnforum/viewthread?thread=1417

I've successfully built an HTTP client using the Zend Platform against Sesame 2.0. Phesame was a very similar project which used PEAR and was only compatible with Sesame 1.x.

Sesame returns JSON it appears, and the Zend platform has a nice decode which works well for that. Still todo is complete SPARQL HTTP POST, and see if the SeRQL has the ORDER BY I need that Sesame doesn't support in SPARQL.

Sesame 2.0 is 2-10 times as fast for the queries I perform than ARC. However, Sesame 2.0 SPARQL doesn't have some features I need, like ORDER BY and full support for OPTIONAL. If these come through soon, I'm going to still need to build a bridge between Drupal/PHP and Sesame. I'll either need:

• Resin, and run PHP inside Java for Sesame
• PHP REST against YARS which uses N3 and no SPARQL
• Redland, with maybe the new lib-b
• PHP REST against Sesame, and just deal with current SPARQL limitations

AHIRC gained a couple of features today via Semantic Search.
Taxonomy weights now sort the display of subcategories, and a hack was completed to do double ORDER BY (see previous post).
Doesn't sound like much, but as they say here in Spain - poco a poco.

RDF stores are not simply returning table rows, the return parts of triplets:
SELECT DISTINCT ?s WHERE{?s ?p ?o. }

You can structure the query to return full "data objects" of sorts:
SELECT DISTINCT ?objectid ?param ?literal WHERE{?objectid rdf:type <[http://example.com/Class#y]>. ?objectid ?param ?literal. }

Then you can ORDER by a specific field:

With a bit more work to get done with AHIRC (work I'm doing with CivicActions and LINC) - I'll need to make Semantic Search fast enough for importing more than 2000 nodes. AHIRC has around 6000 nodes and can take 10-12 seconds for longer searches. Why?

• MySQL 4 had a bug* with the statistics phase of optimizing SQL when you had many inner joins on one large table (this actually caused queries to hang)