Identifying synonymy between SNOMED clinical terms of varying length using distributional analysis of electronic health records.

摘要: Medical terminologies and ontologies are important tools for natural language processing of health record narratives. To account the variability use, synonyms need to be stored in a semantic resource as textual instantiations concept. Developing such resources manually is, however, prohibitively expensive likely result low coverage. facilitate expedite process lexical development, distributional analysis large corpora provides powerful data-driven means (semi-)automatically identifying relations, including synonymy, between terms. In this paper, we demonstrate how corpus electronic records - MIMIC-II database can employed extract SNOMED CT preferred A distinctive feature our method is its ability identify synonymous relations terms varying length.