Expert/consultation system for a retrieval data-base with semantic network of concepts

This paper describes a development and implementation of an expert/consultation system for a retrieval data-base, that interfaces between the user and a retrieval system. The system's objective is to perform the information consultant's job in assisting a user to select the right vocabulary terms for his query. It is particularly useful for a novice user of a controlled-vocabulary, index-based retrieval system, who is not familiar with the vocabulary and the system Thesaurus. The user will enter his terms/keywords, that represent his information need, and the system will apply search procedures on its knowledge-base, and will find relevant concepts to be used as query-terms. The system is interactive; it can explain to the user why/how a concept was discovered/suggested, and it can backtrack and try to find alternatives in case the user rejects a suggested concept. Two versions of the system were developed, utilizing two search and interaction strategies. Experiments will be conducted with the two alternatives in order to find out user preference and to compare performance. Performance will also be compard with an alternative "conventional" approach, which is an On-Line-Thesarus - developed as part of this study.