Modeling of artifacts and ontology

A model of an artifact for a model-based problem solving represents constituents of the target system and principles governing them. The modelers essentially make some assumptions in order to restrict and simplify the models. However, in the conventional model-based problem solving systems, such assumptions are rarely explicit. As a consequence of such implicitness, when one wants to reuse a model, it is difficult to know the model is appropriate for his/her purpose. Ontology plays a crucial role in clarify the assumptions behind the models and the problem solving systems.

The current research aim is to reveal the assumptions about "causality" and "time interval" in the qualitative reasoning (QR) systems. Firstly, we propose "a causal time ontology" to represent conceptual meanings of time intervals (called "causal time scale") among causal relations generated by the QR systems. The ontology aims to enumerate all possible temporal meanings of the models, and to clarify the temporal performance of the QR systems with respect to causal ordering, called "causal time resolutions". In the paper, the performances of some conventional QR systems (e.g., QSIM) are clearly specified using the concepts in the ontology. (see the paper in IJCAI-97).

Moreover, we identified 7 causal time scales for fluid-related systems such as power plants. On the basis of them, we built a domain ontology of the fluid-related systems which provides reusable models of crucial components in the systems. Our qualitative reasoning system can derive more finer-grained causal relations from the reusable component models than the conventional QR systems (see the paper in PRICAI'96)

Major publications:

1. A Causal Time Ontology for Qualitative Reasoning, Proc. of IJCAI-97, 1997
2. A Qualitative Reasoning based on an Ontology of Fluid Systems and Its Evaluation on a Power Plant, Proc. of PRICAI'96, 1996