THE COMPARATOR IDENTIFICATION METHOD FOR DYNAMIC FILLING THE THESAURUS OF OPERATIONAL SEARCH ACTIVITIES.

The use of the thesaurus considerably improves recall and precision of relevant information output of integrated information and criminalistic systems used in operational search activities. The design difficulty of the objectoriented criminal thesaurus consists in the necessity of its constant changes and adapting to new domains. This paper proposes a method for automatic dynamic filling of this particular kind of thesaurus. On the basis of existing thesaurus development standards we propose the semantic linguistic processor for automatic selection of information terms and the method of comparator identification for descriptorization of received notions. The functions of understanding a connected text, key terms, a conceptual-semantic predicate, and a descriptorization predicate introduced in the paper enable to divide the set of terms into mutually exclusive semantic equivalence classes corresponded to semantically similar concepts. This method allows detecting automatically descriptive dictionary entries of a dynamically variable thesaurus adapted to textual information arrays that are used in the process of operational search activities. The method is implemented as an automatic thesaurus of the information retrieval subsystem integrated by information and criminalistic systems. The conducted experimental study of subsystem results has shown quite high indicators of recall and precision of subsystem outputs. [ABSTRACT FROM AUTHOR]