Ontology Population from Software Requirements Specifications using Patterns and Formal Grammar
Ontologies are machine-readable models to represent knowledge in a particular domain. For ontologies to be used in the application of a particular domain, instances and their relations are needed. Ontology population is a process for adding instances and their relations in an ontology. Manual population of ontologies is a time-consuming and challenging task. Therefore, semi-automatic or automatic methods are required for ontology population. This paper presents a semi-automatic process to populate ontologies from a semi-structured English text. The process uses lexico-syntactic patterns, a formal grammar and a seed ontology to parse text and extract information from it to populate the ontology. We have developed a recursive-descent parser to parse text and extract information to create instances and define their relations. The results are evaluated using a software requirements specification for a telecommunication component of a safety-critical system in the avionics domain. Preliminary results are promising, and overall precision and recall are over 85\%. The proposed process does not depend on common natural language processing tasks such as part-of-speech (POS) tagging or Named-Entity Recognition (NER).
Muhammad Ismail – Jönköping University