AKTors.org
AKTors.org AKTTechnologiesPublicationsRelated ProjectsPeople
AKTors.org AcquisitionModellingRetrievalReusePublishingMaintenance
CJJTOE from The University of Aberdeen

 

CJJTOE fact-file

Owner  :  University of Aberdeen
Researchers
(listed alphabetically)
 :  David Corsar, Derek Sleeman
Description  :  http://www.csd.abdn.ac.uk/~dcorsar/software/CJJTOE
Builds on  : 

CLIPS, JESS, JessTab

Addresses challenges  :  Knowledge Acquisition, Knowledge Reuse

What's the Problem?

The CLIPS expert system shell provides a popular environment for building (forward-chaining) expert systems and has been largely re-implemented in Java as JESS, the Java Expert System Shell.  Typical of expert systems, both CLIPS and JESS allow the user to define rules to provide inference over some set of facts, whose structure is defined in a series of templates.  However, both CLIPS and JESS do not explicitly require that templates be defined, and can infer templates based on the structure of facts used in the rules and knowledge base.  JessTab is a plug-in to the popular Protégé ontology editor which links the ontology/knowledge base of Protégé with the JESS inference engine. 

When attempting to reuse CLIPS, JESS and JessTab rule sets it can be helpful to know the structure of the facts used by the rule set.  It can also be useful to have an understanding of the relationships between different types of facts.  As CLIPS, JESS and JessTab programs are written in flat text files, the user is required to search this program code, extract the structure of facts (either defined in templates, or inferred from references to facts in the rules), and then if desired attempt to infer any taxonomic relationships between them.  This can be a very laborious task for substantial rule sets, and inferring the superclass-subclass relationships without any support is a long, error prone task.

Towards a Solution

We have developed a tool which parses CLIPS, JESS and JessTab rule sets, extracts the structure of different facts / templates and attempts to infer any superclass-subclass relationships between them.  It does this by examination of the attributes of each fact template, looking for potential inheritance and suggesting an appropriate superclass-subclass relationship.  The tool produces an OWL Lite ontology containing all the fact templates as classes, structured in the inferred taxonomy.  Further, any facts in the CLIPS, JESS, or JessTab rule set are maintained as instances of the relevant class in the generated ontology.  The tool is currently available as a Java application and web-service.

Screen Shot
CJJTOE

Obtaining the technology
CJJTOE is currently available as a Java program and a web service.  For details, including examples, of how to access and use the web service, see http://www.csd.abdn.ac.uk/~dcorsar/software/CJJTOE.

Technical requirements
CJJTOE is available as both a Java application and as a web service.  The Java version requires a version of Java be installed.  The only technical requirements for the web service are the ability to connect to it. 

Take A Guided Tour
A video is available at http://www.csd.abdn.ac.uk/~dcorsar/software/CJJTOE/cjjtoe-demo.avi

Relevant Links
CLIPS – http://www.ghg.net/clips/CLIPS.html
JESS - http://herzberg.ca.sandia.gov/jess/
JessTab - http://www.ida.liu.se/~her/JessTab/
Protégé – http://protege.stanford.edu