W3C | Submissions

Team Comment on "SPARQL Inferencing Notation (SPIN)" Submission

The W3C is pleased to receive the SPARQL Inferencing Notation submission edited by Holger Knublauch (TopQuandrant), James A. Hendler (RPI), and Kingsley Idehen (OpenLink).

Although not structured along these lines, the technical content of the Submission can be characterized by three, independent though interrelated parts:

  1. There is a general definition to assign rules to classes. These rules are “fired” for instances of those classes, and they can be categorized further as constructors, constraint checkers, or simply general rules.
  2. The Submission identifies SPARQL as a means to define these rules, using SPARQL’s CONSTRUCT and ASK queries, as well as SPARQL 1.1’s INSERT and DELETE
  3. In order to keep the rules assigned to a class as part of the class specifications itself, an RDF encoding of SPARQL (including SPARQL 1.1) is defined in a separate document.

The Submission includes further technical details on creating templates for rules, on accessing a resource instance from a rule, etc. However, these points cover the high level structure of the technical content.

There are notable overlaps between the approach outlined by the Submissions and the results of the RIF Working Group. It should be perfectly possible to use RIF Rules in item #1 above: the definitions of, say, constraints or constructors do not depend on the specificities of SPARQL. To achieve the equivalent to item #3, an RDF encoding of RIF is necessary. An approach to this is being published by the RIF working group, and we would encourage the submittors to consider using this serialization in an updated version of SPIN.

Of course, binding SPIN to RIF instead of restricting to SPARQL does raise a more general question. The RIF Working Group has defined a model-theoretical definition on how RIF and, say, RDF Schema “coexist” in RIF RDF and OWL Compatibility, whereas the submission, effectively, proposes an alternative, more “object-oriented” approach. It will be important to describe the connection between the two approaches so that the possible choices are made clear to the Semantic Web community.

Next Steps

As of today, the W3C RIF Working Group is in a low-activity state, having published the RIF Recommendation. It is not clear at this moment whether the group will take on new work in the future; this will be determined in part by the adoption rate of RIF. If the RIF group does take on new work, SPIN should be one of the inputs.

We encourage people interested by this work to discuss on the semantic-web@w3.org Mailing List [public archive] or public-rif-dev@w3.org Mailing List [public archive]

Ivan Herman, Semantic Web Activity Lead <ivan@w3.org>
Sandro Hawke RIF WG Staff Contact <sandro@w3.org>