The work is devoted to the study of the problems of ontology automating creation of scientific subject areas using methods of automatic analysis of texts in natural language. The aim of the work is developing methods for automatic generation of lexico-syntactic patterns for extracting information and ontology augmentation based on the analysis of content ontology design patterns for scientific knowledge areas developed within the framework of the Semantic Web concept. Ontology design patterns are a structured description of top-level concepts in terms of classes, attributes and relationships, and also include competency questions in natural language that serve for understanding and correctly interpretation the properties and relationships of the concept by users. The article proposes an approach to the generation of lexico-syntactic patterns based on questions of competence assessment. The process of generating lexico-syntactic patterns includes the generation of a subject dictionary, the extraction of ontology entities and the formation of a pattern structure based on the Data Property and Object Property, and the generation of semantic, grammatical and positional constraints. Competency questions are used to identify grammatical and positional restrictions necessary for search for ontological relationships in texts. The ontology "Decision support in weakly formalized areas" and the corpus of scientific texts of the same subject area were used in the experiment. During the experiment, the following results were obtained: the degree of ambiguity of the generated templates - 1.5, the F1-measure of assessing the quality of the search for attributes and relations of objects - the F1-measure was 0.77 for attributes and 0.55 for relations, respectively. Comparison of the results obtained for patterns without grammatical restrictions and the results obtained for patterns with grammatical restrictions showed that the addition of restrictions significantly improves the quality of extraction of ontology objects.