2020/06/06 by Katikapalli Subramanyam Kalyan, Kalyan, Katikapalli Subramanyam, S. Sangeetha +1
Biochemistry, Genetics and Molecular Biology · Computer Science · #Biomedical Text Mining and Ontologies #Computation and Language (cs.CL) #FOS: Computer and information sciences #Text and Document Classification Technologies #Topic Modeling
paper · pdf · doi:10.48550/arxiv.2006.04014
openalex publication_date 2020/06/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Medical concept normalization helps in discovering standard concepts in\nfree-form text i.e., maps health-related mentions to standard concepts in a\nvocabulary. It is much beyond simple string matching and requires a deep\nsemantic understanding of concept mentions. Recent research approach concept\nnormalization as either text classification or text matching. The main drawback\nin existing a) text classification approaches is ignoring valuable target\nconcepts information in learning input concept mention representation b) text\nmatching approach is the need to separately generate target concept embeddings\nwhich is time and resource consuming. Our proposed model overcomes these\ndrawbacks by jointly learning the representations of input concept mention and\ntarget concepts. First, it learns the input concept mention representation\nusing RoBERTa. Second, it finds cosine similarity between embeddings of input\nconcept mention and all the target concepts. Here, embeddings of target\nconcepts are randomly initialized and then updated during training. Finally,\nthe target concept with maximum cosine similarity is assigned to the input\nconcept mention. Our model surpasses all the existing methods across three\nstandard datasets by improving accuracy up to 2.31%.\n