In preparation for Semantic Technology 2010 next week, we made an early release of PelletServer for people to start playing with; and we also refreshed the Pelorus user interface, as well as released our Semantic Search integration for Pelorus.
Check it out: we’ve got three running Pelorus demos:
- Baseball: career statistics of every Major League Baseball player from 1876-2007
- NASA: information about space craft, missions, astronauts from the early space program
- Data.gov: the data.gov catalog of publicly released government datasets
New and notable is the refreshed Pelorus user interface, which we think looks cracking good. Also our Semantic Search service (codenamed “Waldo” for now, but that will change soon) is integrated with Pelorus, which is now a browse and search tool.
Lastly, PelletServer is our new RESTful server that integrates all of our key data and analytic services, providing access to those services from any programming language or environment: our Semantic Web infrastructure isn’t just for Java any more.
You can see a running version of PelletServer online—be warned, of course, this is mostly for programmers, since there’s no HTML admin interface yet. There is a RESTful service description interface that supports JSON and RDF (Turtle and RDF/XML) formats. You can use it like this:
[$]> ; wget --header "Accept:text/turtle" http://ps.clarkparsia.com/ -O ps-desc.ttlPelletServer currently provides RESTful access to Pellet’s OWL reasoning services, including SPARQL-DL query, to our Semantic Search service, as well as the service discovery and a prefix-namespaces mapping service. In the next release we’ll include
- relational machine learning framework that’s developed specifically for OWL and RDF
- Terp SPARQL queries (well, that’s present in the current version, but unadvertised)
- Pellet Integrity Constraint Validation
- full HTML admin interface, including JMX tooling
- initial SPARQL Update support
- HTN Planning system, HotPlanner
After SemTech next week we’ll release the first PelletServer client library (Java), which will be available from Github and licensed as Apache 2.0—one of the joys of REST, of course, is how easy client libraries are to build from the service description alone, so I expect to see some clients from other folks, too. We’ll be doing at least a JavaScript and Scala client at some point.
If you’re interested in this technology, or how you can use it to solve hard problems, shoot me an email or if you’re in San Francisco next week, find me at SemTech.
