北京大学课程推荐引擎的设计和实现  

Design and implementation of the course recommendation engine in Peking University

作　　者：沈苗[1] 来天平[1] 王素美[1] 彭一明[1] 高志同 

机构地区：[1]北京大学计算中心,北京100871

出　　处：《智能系统学报》2015年第3期369-375,共7页CAAI Transactions on Intelligent Systems

基　　金：北京大学教育发展研究中心2013北大研究课题(Y+25)

摘　　要：为了在管理信息系统中向师生提供更个性化、人性化的服务,将推荐引擎应用到北京大学学生选课系统中,设计并实现北京大学课程推荐引擎。改进的推荐算法是一种以学生属性分类为前提的协同过滤算法,通过分析选课系统中课程推荐与商业推荐的不同点和学生属性、改进学生相似度的计算方法实现推荐。以北京大学学生选课系统为平台,2013—2014学年度第1学期的10 682名本科生为测试集,为其推荐课程,推荐结果的准确率为34.6%。该系统为学生选课提供了有效的指导,填补了选课系统智能化、个性化方面的空白。In order to provide a more personalized and humanized service for students and teachers in the management information system,a recommendation engine is applied to the Peking University student course-selecting system. The Peking University course recommendation engine is designed and implemented. The improved recommendation algorithm is a collaborative filtering algorithm on the premises of student attribute classification. The recommendation is achieved through analyzing different points between course recommendation and commercial recommendation,analyzing student attribute and improving the calculation method of students ＇ similarity. As the platform,course-selecting system in Peking University recommend courses to 10 682 undergraduate students in the first semester of the 2013-2014 school year. The precision of recommendation results is 34.6%. This system can provide the effective guidance for the students choosing courses,and fill the blank of the system＇s intelligent and personalized.

关 键 词：北京大学 课程推荐 推荐引擎 协同过滤 选课系统 管理信息系统 

分 类 号：TP391.3[自动化与计算机技术—计算机应用技术]