You are here
Comparing SNOMED CT and the NCI Thesaurus through Semantic Web Technologies
Objective: The objective of this study is to compare two large biomedical terminologies, SNOMED CT and the National Cancer Institute (NCI) Thesaurus, through Semantic Web technologies. Methods: The two terminologies are converted into the Resource Description Framework (RDF) and loaded into a common triple store. The Unified Medical Language System (UMLS) is used to identify correspondences between concepts across terminologies. Concepts common to both terminologies are compared based on shared relations to other concepts. Results: A total of 20,369 pairs of equivalent SNOMED CT and NCI Thesaurus concepts were identified through the UMLS. The highest proportion of shared relata is for the superclasses traversed recursively (75% of the concepts share at least one superclass). Slightly more than half of the concepts studied share at least one associative relation (direct relation or inherited from some ancestor). Conclusions: Overall, SNOMED CT and NCI Thesaurus concepts exhibit a relatively small proportion of shared relata. Semantic Web technologies, including RDF and triple stores, are suitable for comparing large biomedical ontologies, at least from a quantitative perspective.