For your semantic web type application, are you using RDF or a proprietary model for internal representation?

Question

If you were working on a project that stores data for use with the semantic network, do you use RDF or even OWL as an internal data format or do you have your own data model / ontology that you map to RDF for interaction?

If you use RDF, what are your impressions of implementing various things, such as cyclic class hierarchies?

If you are doing your job, how is it different from RDF / OWL?

Answer 1

I work with the Jena team at HP (indeed, contributed to Jena), so using Jena is a pretty easy choice on our team. However, there are more reasons than just the convenience of the “next bench”. Various standards for the semantic network contain a lot of details and difficulties, and getting the right is not easy to do on your own. I came across several downloadable ontologies and other datasets that, for example, do not meet the IRI specification. In a fully standalone application, this probably doesn't matter much if you cut corners off standards, but in this case you need to ask why you use semantic web technology in the first place. For me, a strong value in the semester approach for an application would be data binding and open data relationships, in which case compliance is pretty significant.

Most of my data is in triple storage, but I use custom tables as indexes for frequently asked queries. If you know the query template in advance, a well-indexed table in a good db engine will be hard to beat for universal three-dimensional storage schemaless.

Obviously, one of the disadvantages of using Jena is that it is specific to Java. I use Jena with jruby, but I am looking forward to a good Ruby RDF library (work continues ). I would also like to get more complete RDF / OWL support in Javascript and Flex when we create complex client interfaces.

Yang

Answer 2

I am currently working on some really small projects in this area, and I “mainly” use RDF, although I use a simple URI registry for parsing to avoid loops in the data structure itself. Although, I must say that I am still at the conceptual stage of these projects. In the end, I rely mainly on third-party repositories such as Jena, rdflib, etc.