Using case based reasoning for basis development in intelligent decision systems

The artificial intelligence technique of case based reasoning is proposed to support the initial structuring of decision theoretic models. Case based reasoning obtains an initial problem structure by accessing and manipulating influence diagrams stored in a large knowledge base of cases. A matching process locates all cases which match a specification on influences which are direct, indirect and analogous. The set of matching cases is reduced to the smallest sub-set which contains all influences. This set is then ranked. If this phase does not produce a suitable basis the matched cases and any unmatched influences are merged to produce a new case. An architecture which incorporates case based reasoning with traditional intelligent decision system architectures is described.