Methods to eliminate uncertainty of lexical-morphological analysis in process of knowledge extraction fromnatural language texts

The techniques for improving the lexical-morphological analysis of the semantic-oriented language processor which carries out the extraction of knowledge from the natural language texts are considered. Methods and means to eliminate uncertainties in the allocation of text tokens and sentences, as well as methods of combinatorial and syntactic analysis to eliminate morphological homonymy are proposed. The methods give possibilities to improve the semantic linguistic processor which extracts knowledge structures from natural language texts and restores implicit information.