Semantic Network Analysis and Optimization Path Research on Medical English Terminology

Abstract

Abstract Specialized terms are often longer words in medical English, and how to improve the effective memorization of specialized terms has become a research hotspot in medical English learning. The article proposes a method for creating a semantic network model for medical English terms using CRFs and the ATT-BiLSTM model. Firstly, on the basis of combing the characteristics and methods of medical English terminology composition, the CRF model is introduced to recognize the named entities of medical English terms and combined with the Viterbi decoding algorithm to obtain the lexical annotation result sequence of medical English terms. Secondly, an entity-relationship extraction model for medical English terms is established by combining the attention mechanism with the BiLSTM model, and ELMo vectors are added to improve the extraction efficiency of medical English terms. Finally, a semantic network model of medical English terms was constructed based on the Neo4j database by combining the named entity recognition and entity relationship extraction model, and the semantic network analysis was carried out by the semantic similarity algorithm based on information content. The results show that the CRFs model only needs a 14.2MB training corpus to co-represent the contextual information of medical English terms, and its F1 value is maximally improved by 10.22% on average after adding ELMo vectors to the ATT-BiLSTM model. The mean semantic similarity of medical English terms in the semantic network model is 1.391, and the result of the semantic similarity assessment is 0.803. Relying on the semantic network model can help learners understand the differences between medical English terms, find more similar terms to elucidate the standard terminology, and improve learners’ comprehension of medical English terms.