Structuring Domain Knowledge by Semi-automatic Ontology Construction

摘要

In this paper, we present a case in semi-automatic ontology construction from literature. For this, we concentrate on the articles about autism obtained from the PubMed Central database. Our motivation was to investigate how separate parts of articles, such as titles, abstracts and full texts, influence the constructed ontology. Our results confirm the intuitive expectation that constructing ontologies from abstracts is a rational choice when uncovering the structure of a given scientific field. In addition, when compared to general knowledge of autism, ontology concepts from abstracts show the highest similarity.