混合性焦虑抑郁障碍服务质量情感主题识别研究  

Research on Sentiment-topic Recognition on Service Quality of Mixed Anxiety and Depression Disorder

作　　者：温廷新[1] 徐桂颖 Wen Tingxin;Xu Guiying(School of Business Administration,Liaoning Technical University,Huludao Liaoning 125105)

机构地区：[1]辽宁工程技术大学工商管理学院,辽宁葫芦岛125105

出　　处：《情报探索》2024年第7期10-19,共10页Information Research

基　　金：辽宁省教改项目基金项目“数据库原理及应用课程建设的研究与实践”(项目编号:2021-39)成果之一。

摘　　要：[目的/意义]为识别在线医疗社区中混合性焦虑抑郁障碍患者评论医疗服务质量情感及主题,提出一种基于CNN-BiLSTM和LDA模型的服务质量情感主题识别模型。[方法/过程]首先,构建CNN-BiLSTM模型提取患者评论内外关键特征得到情感倾向分布;其次,运用LDA主题模型提取患者正负向评论主题,结合《医院评价标准(征求意见稿)》得到医疗服务质量主题,从分布和情感词对正负向服务质量进行挖掘。[结果/结论]CNN-BiLSTM的F1值为94.43%,均优于其他对比模型;结合LDA主题模型和相关文献得到5维医疗服务质量主题及分布;根据主题情感词及分布得到负向评论产生的主要原因,为识别和改善医疗服务质量提供有效决策支持。[Purpose/significance]In order to identify the sentiment and topic of medical service quality in the comments of patients with mixed anxiety and depression disorder in the online medical community,a service quality sentiment-topic recognition model based on CNN-BiLSTM and LDA model is proposed.[Method/process]Firstly,the CNN-BiLSTM model was constructed to extract key internal and external features of patients’comments to obtain the distribution of emotional disposition.Secondly,the LDA topic model was used to extract the topics of patients’positive and negative comments.The medical service quality topics were obtained by combining Hospital Evaluation Standards(Draft for Comments),and the positive and negative service quality was mined from the distribution and emotional words.[Result/conclusion]The F1 value of CNN-BiLSTM is 94.43%,which is better than other comparison models.The topics and distribution of 5-dimensional medical service quality were obtained by combining LDA topic model and related literature.The main causes of negative comments were obtained according to the topic sentiment words and their distribution,which provided effective decision support for identifying and improving the quality of medical services.

关 键 词：在线医疗社区 服务质量 混合性焦虑抑郁障碍 情感分析 主题模型 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]