Using neural networks and expert systems to model users in an object-oriented environment

The authors discuss the usefulness of neural networks and expert systems for modeling users in an environment in which information about software is retrieved. User modeling offers an effective technique for determining user requirements and for enhancing the relationship between a user´s cognitive abilities and software sophistication. A hybrid architecture is proposed in which the expert system is used for solving well-defined problem domains, and the inferred results are interpreted. The neural network is applied to recognize the user´s pattern of stereotypes dynamically and solve the reasoning conflicts. To achieve a higher interaction efficiency and simplify the maintenance of the knowledge base with these two heterogeneous reasoning processes, an object-oriented environment is proposed as an integration platform. The framework of a prototype implementation is described which is an object-oriented intelligent consulting system for selecting appropriate problem-solving software