SEMANTIC INTERACTIVE SEARCH MODEL IN DOCUMENTARY INFORMATION-SYSTEMS

The current state of development of interactive search models in documentary information systems characteristically involves the use of the Bayes method of statistical decision theory for modeling. This approach allows the correction, during the course of model training, of the input information needs of the system user as expressed in the request language. The probabilistic-statistical search models, however, are incapable of a semantic modification of requests. The solution of this problem depends on the possibility of finding a way of qualitative evaluation of relations inherent in the information retrieval thesaurus. The model proposed here is based on solution of the following problems: (a) development of a mathematical model of thesaurus; (b) isolating a set of relations adequate for information retrieval (enabling the performance of the possible strategies of request modification) and a qualitative evaluation of these relations; (c) assessment of the semantic strength of terms in a request for inducing the sequence of request modification; and (d) development of algorithms for model use in information retrieval.