基于组合模型的医生推荐系统研究与实现

发布时间：2018-06-05 19:29

本文选题：个性化推荐 + 层次分析法　；参考：《东华大学》2013年硕士论文

【摘要】：个性化推荐是一种智能的主动的为用户提供其可能感兴趣的产品或者服务的互联网新应用,它能够有效的提高用户对产品或者服务供应商的忠诚度,提高企业效益,与此同时,它能减少用户搜寻其可能感兴趣的产品或者服务所花费的时间和精力,为用户提供更加智能的服务。目前个性化推荐已经成功的应用于互联网服务领域,如电子商务、视频点播服务、网络广告投放、新闻订制等。上海医联预约平台是在上海申康医院发展中心开展的医联工程的应用基础上,面向申康所辖市级医院提供包括门户技术平台和预约挂号服务的系统,它也属于一种互联网服务领域,但确少有人研究如何把个性化推荐引入医联预约服务中,并以此来提高医联预约服务质量和用户体验。本文首先对个性化推荐技术进行了较为全面、深入的分析。分别用户建模,推荐对象建模,推荐算法三个方面介绍了个性化推荐系统,介绍了个性化推荐系统的三种实验方法和四个主要的评价指标,重点阐述了主流推荐算法的实现方法和优缺点。接着,针对上海医联预约平台存在的信息过载和“冷热不均”现象,探索了将个性化推荐引入医联预约平台的可能性和意义。本文根据预约平台的数据特征和患者的预约需求,以预约绩效和出诊绩效为指标,使用层次分析法建立了医生绩效模型,然后根据患者预约历史记录的统计结果,结合协同过滤的思想,建立了患者偏好模型。在上述基础上,本文提出了一种基于组合模型的医生推荐算法,并根据实验反馈,进行了相应的改进和优化。最后,本文在上海医联预约平台上,根据设计的医生推荐算法设计了医生推荐系统,并详述了医生推荐系统的数据库设计、离线处理算法的实现、在线推荐算法的实现和界面设计。根据医生推荐系统上线后记录的患者操作日志对推荐系统进行了评测。评测结果表明本文设计的医生推荐系统的推荐结果合理有要,且符合患者的预约需求。
[Abstract]:Personalized recommendation is an intelligent and active new Internet application that provides users with products or services that they may be interested in. It can effectively enhance the loyalty of users to product or service providers and improve the efficiency of enterprises, at the same time, It reduces the amount of time and effort that users spend searching for products or services they may be interested in, and provides more intelligent services for users. At present, personalized recommendation has been successfully applied in the field of Internet services, such as e-commerce, video-on-demand services, online advertising, news customization and so on. On the basis of the application of the Medical Union Project carried out by the Shanghai Shenhang Hospital Development Center, the Shanghai Medical Association booking platform provides a system for the municipal hospitals under the jurisdiction of Shankang, which includes a portal technology platform and an appointment registration service. It is also a kind of Internet service field, but few people study how to introduce personalized recommendation into Medical Union reservation service to improve the service quality and user experience. In this paper, the personalized recommendation technology is analyzed comprehensively and deeply. In this paper, user modeling, recommendation object modeling and recommendation algorithm are introduced respectively, and three experimental methods and four main evaluation indexes of personalized recommendation system are introduced. The implementation method, advantages and disadvantages of the mainstream recommendation algorithm are expounded. Then, in view of the information overload and the phenomenon of "uneven cold and heat", the possibility and significance of introducing personalized recommendation into the platform are explored. According to the data characteristics of the booking platform and the patient's reservation demand, this paper establishes the doctor's performance model by using the AHP, and then according to the statistical results of the patient's appointment history record, the author takes the appointment performance and the visiting performance as the index. Combined with the idea of collaborative filtering, a patient preference model was established. On the basis of the above, this paper proposes a doctor recommendation algorithm based on combinatorial model, and improves and optimizes the algorithm according to the experimental feedback. Finally, this paper designs the doctor recommendation system on the platform of Shanghai Medical Association, and describes the database design of the doctor recommendation system and the realization of off-line processing algorithm. The realization and interface design of online recommendation algorithm. The recommendation system was evaluated according to the patient operation log recorded after the introduction of the doctor recommendation system. The evaluation results show that the recommended results of the doctor recommendation system designed in this paper are reasonable and meet the needs of patients.
【学位授予单位】：东华大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：TP311.52

【参考文献】