An Ontology-Based Human Friendly Message Generation
in a Multiagent Human Media System for Oil Refinery Plant Operation

Riichiro Mizoguchi, Toshinobu Sano and Yoshinobu Kitamura

ISIR, Osaka University
{miz, sanop, kita}@ei.sanken.osaka-u.ac.jp

This paper mainly describes the role of knowledge media in the project of “Development of a human interface for the next generation plant operation” running as a subproject of Human Media project funded by MITI. The role of an ontology and ontology server in the interface for oil-refinery plant operation is described. They play a central role as the Knowledge media which is one of the key technologies in Human Media project. An oil-refinery plant ontology as well as an operation task ontology have been developed. Ontology server, which stores and provides a plant ontology, has two major functions in the system:

(1) To standardize each agent’s understanding of the target world, the oil-refinery plant in our case, by providing a plant ontology.

(2) To generate human friendly messages in terms of appropriate words operators use daily.

The latter is the main task of ontology server in the current implementation which is done by focused point tracing module and adaptive word selection module both of which consult the plant model built using the components obtained by instantiating the concepts in the plant ontology. The system has been implemented and preliminary evaluation has been done successfully.