Inter Concept Ontology Similarity-Mapping of Feature Based Similarity to Information Theoretic Domain

Semantic similarity mechanism is mandatory in information retrieval, information integration, ontology mapping and psycholinguistics. Mapping from feature based model to information theoretic domain has been proposed to find the semantic closeness of concepts belonging to different biomedical ontologies. This project also aims in adapting existing intra ontology feature based approach (Pirro) and propose methods for assessing semantic similarity among concepts from inter ontologies. The possibility of extending Dice coefficient measure to inter concept ontology similarity is also attempted. The proposed idea was tested using MeSH and SNOMED-CT biomedical ontologies. The study of using Dice coefficient in information retrieval was also proposed. The proposed approach is corpus independent and it correlates well with the human judgements. The correctness of the proposed similarity metric is proved by comparing the result against the human judgements by computing the correlation coefficient. The proposed approaches outperform the other path length based computational methods as it achieves the highest correlation.