[目的/意義]基于在線醫療評論數據識別醫生服務評價主題,為醫療服務質量管理提供參考。[方法/過程]基于框架語義理論建立細粒度情感語義分類詞典,基于框架和語義角色構建醫生服務質量評價主題知識庫;對評論文本進行框架語義標注,結合語義相似度算法,將文本中的語義信息與評價主題知識庫相匹配,從而識別在線評論中的評價主題。[結果/結論]通過醫療評論數據測試,獲得了較高的正確率和召回率,證明了研究方法的有效性和科學性,是情感分析在醫療健康領域的一次有益探索。

[Purpose/significance] Based on the online medical review data,topics of physician service evaluation are identified to provide reference for medical service quality management.[Method/process] Based on the Frame semantics,a fine-grained emotion semantic classification dictionary is established.Based on frames and semantic roles,the knowledge base of physician service quality evaluation topics is constructed.At last,the semantic information of the review texts is matched with the evaluation topics knowledge base and thus the evaluation topics are identified after the Frame semantic annotation of the texts,combined with the semantic similarity algorithm.[Result/conclusion] Through tests of medical review data,the high correct rate and recall rate are obtained,which proves the validity and scientificity of the research method.It is a useful exploration of sentiment analysis in the field of medical health.