When describing inference structures by IPST, the knowledge that is used there must be also described. Because knowledge bases have been described in different vocabularies at different sites, expert systems need to correspond their knowledge to knowledge in other expert systems. Because this paper's main objective is not to approach the issues regarding the correspondence, a preliminary and naive approach has been taken by using a machine-readable dictionary called EDR concept dictionary (Japan Electronic Dictionary Research Institute LTD.,1993). It contains about 400,000 concepts, and concept meanings have been defined using such concept relation labels as agent, goal, cause, etc. Therefore it looks like a huge and complicated semantic network. For the later experiment, a common domain ontology has been constructed manually by examining where all concepts from the task-domains are located in EDR concept dictionary. Thus the common domain ontology comes up as shown in Figure 2.
Figure 2: A Common Domain Ontology Manually Constructed