Логико-лингвистическая модель генерации фактов из текстовых потоков информационной корпоративной системы

Abstract

This paper proposes a logical-linguistic model extracting semi-structured facts in English texts. To identify the fact some entities expressed by lexical units as well as semantic relations between them are defined in the text. The semantic relations are expressed by semantic functions of sentence participants. A fact is written in form of a triplet: Subject – Predicate – Object, in which the Predicate represents the relations and Subject and Object define the subjects, objects or concepts. Two types of the facts are defined. The first type is fact that describes relation between two entities; the second one is fact that fixes the value of a predetermined attribute. The functions are described by predicates of algebra of finite predicates. The mathematical model allows associating meaning relations of concepts of a sentence with elements of the syntactic and morphological structure of the English sentence. The model is applied to the semantic stage of linguistic processor of information subsystem for facts identification, which are essential for business analysis, in the framework a semi-structured texts. Software implementation of the model is designed. The input subsystem receives text streams disparate sources of information of the integrated corporate system, basic facts of space of the system are output. The accuracy and completeness extracted facts from texts in English by the subsystem are compatible with extracted facts by an expert.