基于CEP的校园推荐系统设计与实现  

Design and Implementation of Campus Recommendation System Based on CEP

作　　者：周苾 卢山[3] 汤鲲 ZHOU Bi;LU Shan;TANG Kun(Wuhan Research Institute,Wuhan 430000,China;Nanjing Signal Fire Software Technology Co.,Ltd,Nanjing 210000,China;Experimental Teaching Center of Computer,Southeast University,Nanjing 210000,Chin)

机构地区：[1]武汉邮电科学研究院,湖北武汉430000 [2]南京烽火软件科技有限公司,江苏南京210000 [3]东南大学计算机教学实验中心,江苏南京210000

出　　处：《计算机技术与发展》2018年第6期192-196,共5页Computer Technology and Development

基　　金：江苏省科技支撑计划项目(2015BAK20B03)

摘　　要：针对高校学生课外资源过载、缺少实时个性化推荐等问题,将复杂事件处理技术(CEP)运用到推荐系统中,发挥其强大的实时处理优势,对进一步提高高校个性化资源推荐系统的准确性和实时性进行了研究。将高校管理与资源推荐相结合,设计大数据背景下的校园学辅资源推荐系统。利用复杂事件处理技术,将学生实时地理位置信息、签到信息、图书馆借阅信息、宿舍信息等四种多维、异构数据源相结合,使用EPL语言实现相应的规则关联,将简单事件流通过Esper引擎处理后形成复杂事件流,对高校学生资源推荐系统作实证分析,实现从学生基本信息数据流处理、复杂事件规则验证到相关资源推荐的整个推送过程。实验结果表明,将该系统与最常用的基于协同过滤算法的推荐系统性能作比较,实时性提升了20%,准确度提升了30%,验证了该系统具有良好的推荐效果。To solve the problems of the overload of extracurricular resources and lack of real-time personalized recommendation for college students,the complex event processing（ CEP） is applied to recommender systems owing to its powerful real-time processing advantages for research on further improvement of the accuracy and real-time of personalized resource recommendation system in colleges.Combining university management with recommending resource,we design the campus resource recommendation system under the background of big data. Using complex event processing techniques to combine four different kinds of multidimensional,isomerism data source,real time location,check-ins,library lending and dormitory information,we use the EPL to change simple events to complex events through the Esper engine,which can implement the whole process from student basic information data flowprocessing,complex event rule verification to related resource recommendation. Compared with collaborative filtering algorithm based on cosine,the results showthat the real-time is increased by 20%,and the accuracy is increased by 30%.It is proved that the recommendation system performs well.

关 键 词：复杂事件处理 个性化推荐 资源过载 协同过滤 Esper 

分 类 号：TP302[自动化与计算机技术—计算机系统结构]