RACER
Racer is enormously powerful, and it's very tempting to use it for Semantic Web projects. Firstly, it's a full inference system which reads OWL (and most systems are RDF-centric, even if they'llmove to the use of OWL in the future). Secondly, it's network-accessible, with load balancing etc. Thirdly it supports a publish/subscribe interface, so you can easily do persistent queries. Fourthly, it has Lisp, Java, C++, TCP, and HTTP interfaces.
The downsides: no explicit Python interface, it's not free, and it can't save as OWL: import only, then you can save the KB in Racer's own format. I'd also have to write my own query interface: the options are (a) a TCP API and (b) DIG over HTTP. Given that all of the existing Semantic Web query interfaces are for RDF (e.g. RDQL, QuizRDF, RQL, SquishQL), ignore inference (operating only on the triples that are there in the KB), aren't persistent, and are in Java it's not a big problem — I'd have to do it anyway.
I guess I'll write my own OWL-QL to Racer transformer. It has to be two-way, too — no point understanding a query if we can't answer in the same language!
In The Evening by Led Zeppelin is a great song. D'Yer Mak'er isn't, but you can't win all the time.
Posted at 2004-03-26 04:35:10 by Richard • Link to RACER
Comments, trackbacks.
