Petrasova S. Information technology of knowledge identification in scientometric systems based on intelligent analysis of weakly formalized data

The object of the study is the process of knowledge identification in weakly formalized information of scientometric systems. The aim of the thesis is to increase the effectiveness of knowledge identification in scientometric systems by designing the models and methods of intelligent analysis of weakly formalized data. Research methods: fundamentals of the theory of intelligence and the finite predicate algebra, the method for component analysis and tools of regular expressions, the method for comparator identification, methods of mathematical statistics. The main results: the logical-linguistic model of semantically connected fragments identification in weakly formalized abstract information has been developed. The model is based on the use of algebraic-predicate operations that allows improving the information technology of knowledge identification in scientometric systems. The method for comparator identification of semantic fragments of abstracts has got the further development. This method is used to classify abstract fragments in scientometric systems that allows determining common information spaces of scientific interaction by modelling intelligence functions of sense classification of abstracts. The method for the formalization of semantic relations between entities has been improved. The method is based on the use of the semantic similarity measure and intelligent analysis for the identification of equivalence and tolerance classes that allows defining implicit relations of similarity and relations of taxonomy. The information technology of knowledge identification in scientometric systems has been improved. The technology allows identifying common research fronts by defining dynamically implicit connections between abstracts of scientometric systems. The research results have been implemented in the systems of summaries and abstracts processing in scientific libraries. Using the developed information technology improves the effectiveness of knowledge identification in weakly formalized data by increasing the average values of the precision and recall measures of semantic similarity of text information.