基于经食管超声心动图记录和基于项目协同过滤的混合推
荐方法
刘星星;王晓东;姚宇
【期刊名称】《计算机应用》 【年(卷),期】2017(037)0z1
【摘要】In order to enhance the clinical application value of TransEsophageal Echocardiography (TEE) ultrasound records and improve the working efficiency of physician users,a hybrid recommendation method based on TEE records and Item-based Collaborating Filtering (ItemCF) was proposed.The features of TEE ultrasound records and the interactions of physician users were combined to achieve the personalized case system recommendation.At first,the data sets were collected from the case query system platform of TEE and preprocessing was done on them afterwards,including the cleaning and filtering of data sets with both TEE records and physician users' interactions,label marking on which could be as recommendation factors for the records and feature extraction.Later on,according to the different query habits of physicians,the interactions of users on TEE records were analyzed,and the hybrid recommendation method of content-based on TEE records and ItemCF was utilized to train the model.Meanwhile,the
recommendation
results
were
updated
dynamically based on the feedback of users and the changes of TEE
records.After the tests of experiments,a more stable model is formed,and the recommendation accuracy is higher when the similarity number is 50 and the recommendation number is 10.%为提升经食管超声心动图(TEE)超声记录的临床应用价值,提高医师的工作效率,提出基于TEE超声记录和基于项目协同过滤(ItemCF)的混合推荐方法,结合TEE超声记录特征和医师用户交互记录实现病例系统的个性化推荐.首先依托TEE病例查询系统平台采集数据并进行预处理,包括TEE记录数据和医师用户记录数据的清洗,标出记录表中可能被作为推荐因素的项目并提取特征.然后针对不同医师查询习惯,分析医师-TEE记录交互数据,运用基于内容和ItemCF的混合推荐,训练模型,同时根据医师反馈信息和TEE记录数据集的变化,动态更新推荐结果.经实验测试,形成了较稳定的推荐模型,并且在相似记录数目为50、推荐记录数目为10的情况下,推荐精度较高.
【总页数】4页(300-302,311)
【关键词】TEE超声记录;用户交互记录;病例查询;基于内容推荐;ItemCF推荐 【作者】刘星星;王晓东;姚宇
【作者单位】中国科学院成都计算机应用研究所,成都610041;中国科学院大学,北京100049;中国科学院成都计算机应用研究所,成都610041;中国科学院成都计算机应用研究所,成都610041 【正文语种】中文 【中图分类】TP391 【文献来源】
https://www.zhangqiaokeyan.com/academic-journal-cn_journal-
computer-applications_thesis/0201241672559.html 【相关文献】
1.基于项目评分预测的混合式协同过滤推荐 [J], 盈艳; 曹妍; 牟向伟 2.基于高斯pLSA模型与项目的协同过滤混合推荐 [J], 陈登科; 孔繁胜 3.基于用户特征和项目属性的混合协同过滤推荐 [J], 廖新考
4.一种基于农产品推荐的混合协同过滤方法的设计 [J], 余明艳; 支和才 5.基于Tag和协同过滤的混合推荐方法 [J], 王卫平; 王金辉
以上内容为文献基本信息，获取文献全文请下载
基于经食管超声心动图记录和基于项目协同过滤的混合推荐方法
