wiki:minutes_955-07

Version 4 (modified by onardmejino, 10 years ago) (diff)

--

Meeting minutes from 9/05/2007

Attendance:

  • Jim Brinkley, Dan Cook, Linda Shapiro, John Gennari, Onard Mejino, Todd Detwiler, Joshua Franklin, Max Neal, Marianne Shaw, Jeff Zavaleta

Summary:

  1. Marianne reported that they (with Dan S and Todd) have been considering adding Datalog extensions to SPARQL to provide the recursion that they are going to need for the queries they want to support. After looking at available implementations, they realized that they would not get the performance benefits that they are going to need for the FMA. Instead, they are looking at how to modify SPARQL and the Jena framework so that they can push more complex queries down to the SQL database. They need to be able to add support for nested queries and for recursion, neither of which is supported by SPARQL.
  • Neither PostgreSQL nor MySQL have support for recursion; the commercial databases (IBM db2, SQL Server 2005, Oracle) do have this support. For now, they are migrating to SQL Server 2005 as the database cluster machines already have it installed.
  • They had been working on optimizing transitive closure for the FMA in the PostgreSQL database using stored procedures. After implementing the semi-naive algorithm, they were able to calculate the transitive closure for all 139 relations in just under 8 hours. After cleaning up the script, Marianne will be going to pass it on to Todd and Joshua.
  1. Todd gave a brief survey, including examples, of the current state of rule languages for semantic web data models (like RDF, RDFS, and the various OWL forms). The 3 that he discussed were: 1. The SWRL tab in Protege with its Jess bridge, 2. A prototype SWRL engine on top of Sesame which converts SWRL rules to SeRQL queries, and 3. Jena rule support. The working demo was over the Radlex-FMA OWL ontology, in main memory, using the Jena rule API. He showed an example of using rules to create transitive closure relationships.
  1. Dan C and John G are travelling to Stanford next Wednesday to meet with Russ Altman and colleagues. They will give a presentation to the group regarding computational application models and the OPB.