This paper proposes a recommendation method of medical literature for clinical diagnostic decision supports.We get all the normalized keywords from the keywords lists in articles,Mesh lexicon and acronyms,then the full relation of keywords and articles are built with normalized keywords.The ideas and detailed processes are also introduced in this paper.Based on keywords co-occurrence analysis,the collections of target sentences and diagnosis-related keywords are abstracted according to existing description of symptoms.And the paper also lays out all the detailed process of measuring keywords and relevant articles with link analysis based on keywords co-occurrence.Finally,we report some related experiments and the results of user evaluations.%本文提出了一种面向临床诊断决策支持服务的医学文献推荐方法,该方法首先对文献关键词从关键词列表、Mesh标准医学词库和缩略词三个方面进行了规范化处理,并据此得到关键词和文献的完整对应关系.文章对相关处理方法的思路和细节都做了详细的说明.然后,利用已知的患者症状描述信息,根据关键词共现形式来获取目标句子集合和诊断相关关键词集合,并利用基于关键词共现形成的关键词链接网络,本文设计了一种测度重要关键词及其相关文献的查询方法.最后,文章对相关实验效果及其用户满意度评价都做了必要的说明.
展开▼
