Extending Disease Ontology with Newly Evaluated Terms to Improve Semantic Medical Information Retrieval

摘要

Semantic similarity plays a vital role in obtaining information resources relevant to an intent of a user query for medical information retrieval. Evaluating semantic relatedness using medical knowledge resources is a problem with constantly arriving new keywords. There is an emerging need to enrich the valuable knowledge resource of Biomedical Ontology with new keywords to enhance the semantic similarity analysis. The existing approaches extend the ontology through identifying a new relationship and senses of the existing terms of the ontology in a manual or semi-automatic operation. The process of manually updating and maintaining the ontology is a time-consuming, labor-intensive and error-prone process, given the dynamic and expanding nature of the biomedical knowledge. Hence, to enhance the semantic similarity measure between the medical terms, the proposed system exploits the sources of disease concepts captured across biomedical resources and automatically identifies the medical terms to extend the Disease Ontology. The Extending disease oNtology witH newly evAluated terms to improve semaNtiC mEdical information retrieval (ENHANCE) system extracts the new disease term, and its relative symptoms exist in the biomedical resource. It associates the extracted terms with its correct ontological class to extend the Disease Ontology. To facilitate this process, initially the system categorizes the Disease Ontology according to the human anatomy and the diseases also classified depending on the anatomical regions in which they originate. Consequently, the terms of the biomedical resource are matched with the anatomy ontologies using vectorization to determine the related anatomy ontology. Thus, the input terms are matched with one or more ontologies, and hence, the system exploits the linkage distance metric that extracts the primary class of the anatomy ontology need to be enriched. After identifying the main class, the name of the new disease and its symptoms are determined through the use of Disease Ontology and the WordNet. With the proposed extended Disease Ontology, the semantic similarity measure enhances the performance of medical information retrieval. The experimental results prove that the ENHANCE system provides an improved biomedical document retrieval system with the proposed linkage distance based semantic similarity measure on the extended Disease Ontology.