The goal is to build a web application that operates on a domain model that's given as an ontology. Our initial attempts at modeling the domain indicate that we'll probably need to work in OWL-Full or just raw RDF. We're looking for a tool chain, ideally one that's free and open, with the following layers:
1) A highly scalable triple store.
So far we've tried a Virtuoso/SuRF stack and a 4Store/ORDF stack. We can't identify a good OWL-Full editor to support either. (Protege is OWL-DL only.) We don't mind changing the stack if there are better options available.
Thanks in advance!
I'll try to keep a running list of various tool chains I've tried and my experience with each so that people don't have to go down the same blind alleys. You can find a structured summary on SemanticWeb.org. Please contribute to it. My own editorial follows:
The format is (triple store) / (orm) / (editor) :
Virtuoso / SuRF / Ontowiki
4Store / ORDF / ?
AllegroGraph / SuRF / TopBraid Composer
? / ? / Protege
? / ? / NeOn
I believe TopBraid Suite satisfies all three, although the criteria for the object-relational mapper may be a bit unclear. TopBraid includes a D2RQ server and a mapping wizard to get you started. D2RQ will map Relational back-ends to RDF/RDFS/OWL in a read-only capacity. If that's what you're looking for then it's a trifecta.
For scalable triple stores, TopBraid includes Jena TDB and SDB. TDB is a native RDF store and SDB maps triples to a relational back-end. TopBraid also plays nicely with AllegroGraph, Sesame, and Oracle RDF. AllegroGraph and Oracle are not open source, although it may be best to see the triple store as doing just that (storing triples) and writing features around it. I.e. view the triple store as the blackest of boxes in your tool chain.
In terms of a "graphical editor", I take this to mean a GUI editor for RDF/RDFS/OWL, including OWL 2. Others have already stated how Composer and Protégé compare ;-). More to the point, though, TopBraid Composer is a full-featured RDF editor, and therefore works with OWL Full. Composer is just the editor, and you may want to look at Live, Ensemble, and Enterprise Vocabulary Network (EVN) to see how semantic application development is supported by TopBraid Suite. TopBraid has SPARQL and OWL-based tools for scripting solutions on top of RDF/RDFS/SPARQL/OWL, and hence works over triple stores.
If you move the "open" criteria from the triple store to the editor/browser/framework, this may do what you need. However, except for TopBraid Composer-Free, which is just the GUI editor (and some import/export features), the platform is not free.
answered 06 Jan '11, 02:19
This is the stack used by oreilly.com for RDF. It's mostly open source.
Tenuki (Jena SDB or Jena TDB based) https://github.com/norcalrdf/Tenuki
Fuseki (Jena SDB or Tena TDB based) http://openjena.org/wiki/Fuseki
Either can be hooked up to Pellet (Open Source but AGPL, not sure of your application)
rdflib... doesn't do much in the way of pythonic access, have not been very happy with it.
https://github.com/norcalrdf/pymantic Our own ORMish wrapper on rdflib, now also working on replacing rdflib itself. The public version is still in early development but we would love feedback. The earlier private version runs http://oreilly.com/catalog/
TopBraid as mentioned by other folks is a great tool for Ontology design. If it doesn't work with Fuseki and Tenuki (which it should, as both are "just" SPARQL 1.1 servers) that should get fixed.
answered 11 Jan '11, 05:23
Can you say why you believe that? I've been using both together. If you have issues please file a bug report (I'm a contributor to SuRF).
answered 14 Jan '11, 09:58