Effectiveness of UMLS Semantic Network as a Seed Ontology for Medical Domain Ontology Building

In the Semantic Web, ontologies play an important role in knowledge portal applications especially in the medical field. Of interest in this project is the manual information extraction process known as the Knowledge Engineering approach to identify relationship pairs of “Colon Cancer Treatment”. The concept and semantic types of these relationship terms extracted are determined using the Unified Medical Language System (UMLS) which is used as a seed ontology. The manual information extraction process yielded some interesting patterns that could further improve the relevance of the Information Extraction (IE) systems. Protégé was used to build and extend the ontology or knowledge base. From this study, it was found that the semantic relationship extracted and compared against UMLS semantic network, only for half of the relationships holds in UMLS. Some semantic types mapped as classes were too general and subclasses were proposed to make it more specific.