We consider the problem of how to represent domain knowledge for user-modelling applications. The problem yields insight into methodology for representing knowledge in hypothetical reasoning systems. In general, there are not enough constraints in a domain to uniquely determine the approach that a reasoning system should take in formalizing its characteristics, but domain-independent methodological principles can help guide the designer. We present the shared information constraint on the domain of inter-agent communication, which implies an interesting representational duality between speakers (designers of utterances) and hearers (recognizers of mental state) in a simple communication environment. This work provides insight into default reasoning applications development.
展开▼