An Approach to Extracting Ontology Concepts from Requirements

Abstract

AbstractThe paper proposes an approach to extracting ontology concepts from the results of automatic processing the software requirements written in natural Russian-language. Relevance of developing an approach to automating the process of extraction of the ontology concepts from the requirements written in natural language is due to the necessity to maintain the requirements specification in a consistent state under conditions of business environment variability. First we discuss the advantages of the ontology-based approach to the requirements engineering as a stage of the software engineering process. The main attention is focused on the possibilities of presenting knowledge about the software application domain and requirements specification in the form of ontologies. We analyzed possibilities of automatic Russian text processing tools for extracting ontology concepts and consider such tools as ETAP-4, MaltParser and UDPipe. The choice of UDPipe as a tool for automatic processing the requirements texts is explained by the best results that it showed when analyzing texts such as software requirements. Then we describe the main classes, object properties and data properties of the proposed requirements ontology and the application domain ontology. An additional advantage of the proposed approach, increasing its practical utility, is building the system of production rules by which the analysis of the results of the automatic processing of the requirements texts obtained using the UDPipe tool.KeywordsRequirements engineeringOntologyAutomatic text processing