基于改进 BTM 模型的医疗服务质量因素识别

Aiming at short text and sparse semantics of online medical reviews, an improved biterm topic model (BTM) topic mining model was proposed based on word co-occurrence analysis (COA) for online medical reviews. Due to the lack of semantic relevance consideration when BTM topic model was used to select word pairs in short texts, a word co-occurrence analysis method was introduced to calculate the semantic relevance, and thresholds were set to screen the participating word pairs for topic mining. Comparing with the traditional BTM and LDA topic models in the topic consistency TC value and JS divergence, the effect of improved COA-BTM was put up in medical review mining. The experiment results show that the improved COA-BTM model can provide a better result in topic consistency and topic quality, proving its effectiveness in the field of online medical review mining. Based on the mining results of this algorithm and SERVQUAL model, the medical service quality factors can be identified more comprehensively.