Machine Learning Selection of Candidate Ontologies for Automatic Extraction of Context Words and Axioms from Ontology Corpus

Abstract

Ontology selection for reuse is a challenging task an ontology/knowledge engineer can face due to the diversity and abundance of ontologies on the internet today. This has become an active research area in ontology engineering. In this study, candidate ontologies are selected from a corpus using Skip-gram, a neural networks algorithm. The word embeddings of the vocabulary of 12 ontologies in the corpus were used to train the Skip-gram models. Thereafter, concepts from a text corpus were utilized to test the models and select the candidate ontologies for context words and axioms extraction. The results showed that the proposed method can automatically extract context words, and the URIs and labels of different types of axioms from the candidate ontologies. This gives the knowledge/ontology engineers detailed information about the contents of candidate ontologies so that they can make an informed decision on the parts of the ontologies they would like to reuse.