Lytvynenko L. Models and methods of analytico-synthetic processing of multilingual textual information in a knowledge-based machine translation system

This thesis is devoted to an actual scientific task - further improvement of models and methods of analytico-synthetic processing of multilingual textual information in a knowledge-based machine translation system. Method of formalization, extraction and analytico-synthetic processing of knowledge hidden in texts written in a natural language was further improved in a way that all the information hidden on the pre-morphological (semiotiсal) level of a text representation is took in account on all the other levels of text analysis. This helps to better extract such important things as: semantically correct sentence boundaries; own names (especially important when they collide with the common vocabulary). Developed a method of automatic syntax analysis of texts for the use in the machine translation tasks which allows to express syntactical rules in a same format for the different languages. Also, this method allows to detect interphase syntax connections with the help of interphase syntax rules which helps to better hold semantical integrity of the text.