RIF Implementation Report

Dear RIF working group,

I am just writing to inform you that we are developing an open-source
RIF-BLD Java implementation and hope to release a first version of our
software at the end of this year. We will publish the code with a liberal
open-source license and we would like to be added to the list of
implementations on the RIF wiki.

The software will be delivered in two components:

1. RIF/XML parser/serialiser - that translates RIF XML documents to a Java
object model (that can be programmatically modified) and a serialiser that
converts back to RIF/XML format.

2. Reasoner components - this translates the Java object model to another
representation suitable for the reasoner (the reasoner components used are
from the IRIS set of reasoning components).

To answer the points requested on the submissions wiki page:

1. Adrian Marte, STI Innsbruck (www.sti2.at) and others (Barry Bishop,
Christoph Fuchs, Matthias Pressnig, Daniel Winkler)

2. The two components are the parser/serialiser/object model called "RIF4J"
and the reasoner, which  will consist of enhancements to
existing software components (WSMO4J, IRIS, WSML2Reasoner) which can simply
be called "IRIS". For a one sentence description: "Parsing and translation
code on top of the IRIS rule-engine". The software will both consume and
produce RIF rule sets.

3. We will support RIF-BLD - which matches with our existing Datalog
rule-engine, although we have extended this with new data-types and
built-ins.

4. We plan to make this implementation conformant.

5. RIF RDF and OWL compatibility has not been addressed yet.

6. We implement some at-risk features, including "equality in rule
conclusion".

7. We would encourage early standardisation on RIF. The most confusing
aspect to RIF so far has been the XML syntax and its inconsistent use of XML
element names, both in terms of upper/lower case and abbreviations.

Regards,
Adrian Marte

Received on Tuesday, 10 November 2009 13:18:13 UTC