DEPARTAMENTO DE INFORMÁTICA Automated Extension of Biomedical Ontologies

Developing and extending a biomedical ontology is a very demanding process, particularly because biomedical knowledge is diverse, com-plex and continuously changing and growing. Existing automated and semi-automated techniques are not tailored to handling the is-sues in extending biomedical ontologies. This thesis advances the state of the art in semi-automated ontol-ogy extension by presenting a framework as well as methods and methodologies for automating ontology extension specifically designed to address the features of biomedical ontologies.The overall strategy is based on first predicting the areas of the ontology that are in need of extension and then applying ontology learning and ontology matching techniques to extend them. A novel machine learning approach for predicting these areas based on features of past ontology versions was developed and successfully applied to the Gene Ontology. Methods and techniques were also specifically designed for matching biomedi-cal ontologies and retrieving relevant biomedical concepts from text, which were shown to be successful in several applications