基于协同过滤改进算法的个性化选课推荐的研究

发布时间：2018-07-17 18:25

【摘要】：随着信息技术的发展，高等院校的教学管理体制发生了相应的变化。我国颁布的《中共中央关于教育体制改革的决定》明确指出：要增加选修课，减少必修课，实行学分制教学和双学位制，而选课是学分制实施的重要环节。各高校的教学管理实践中，，为学生提供了大量的选修课程，门数众多，并且存在课程门类、专业等结构性的不足和缺憾，诸如课程资源中心的组织和管理方式及现今多数的选课方式下，学生难以选择到适合的、符合个人专业发展及个性需求的课程。目前国内不少高校实际实施的是不完全学分制，学生选课的余地较小。鉴于此，我们将个性化推荐技术应用到选课系统中，根据学生的学习需求和兴趣偏好，为学生提供合理、科学和个性的选课推荐，从而避免学生选课的盲目性，提高课程资源的利用率和选课质量。本文主要深入研究协同过滤技术和选课推荐系统的设计，其研究工作如下：首先对个性化推荐技术的优缺点进行分析，提出了基于课程属性和属性值偏好矩阵的协同过滤改进算法。对于协同过滤算法的数据稀疏和冷启动问题，采用课程特征属性和属性值偏好矩阵来加以解决，并采用离线方式计算相似度，从而实现课程的实时推荐。其次针对如何合理的分配项目间的推荐比例，本文将构建一个包含个性化推荐、排行榜推荐和新课程推荐三大模块的系统架构。利用课程属性和属性值偏好矩阵的协同过滤改进算法架构具有个性化推荐功能的选课系统，同时要能减少课程推荐的误差，提高推荐的实时性，从而扩展学生的视野、提高学生学习的自主性和培养学生的创新型思维。
[Abstract]:With the development of information technology, the teaching management system of colleges and universities has changed accordingly. The decision of the CPC Central Committee on the Reform of the Education system promulgated by our country clearly points out that the elective courses should be increased, the required courses should be reduced, the credit system teaching and the double degree system should be implemented, and the elective course is an important link in the implementation of the credit system. In the practice of teaching management in colleges and universities, a large number of elective courses have been offered to students. Such as the organization and management of the curriculum resource center and most of the methods of selecting courses nowadays, it is difficult for students to choose suitable courses that meet the needs of individual professional development and personality. At present, many colleges and universities in China actually implement the incomplete credit system, and the students have less leeway to choose courses. In view of this, we apply the personalized recommendation technology to the course selection system, according to the students' learning needs and interest preferences, to provide students with reasonable, scientific and individual course selection recommendations, so as to avoid the blindness of students' choice of courses. Improve the utilization of curriculum resources and the quality of course selection. This paper mainly studies the collaborative filtering technology and the design of the course selection recommendation system. The research work is as follows: firstly, the advantages and disadvantages of the personalized recommendation technology are analyzed. An improved collaborative filtering algorithm based on curriculum attributes and attribute preference matrix is proposed. For the problem of data sparsity and cold start of collaborative filtering algorithm, curriculum characteristic attribute and attribute value preference matrix are used to solve the problem, and offline method is used to calculate the similarity so as to realize the real-time course recommendation. Secondly, in view of how to allocate the recommended proportion among items reasonably, this paper will construct a system architecture which includes three modules: personalized recommendation, ranking recommendation and new curriculum recommendation. Based on the collaborative filtering of curriculum attribute and attribute preference matrix, the course selection system with personalized recommendation function should be constructed. At the same time, the error of course recommendation should be reduced, and the real-time performance of course recommendation should be improved, so as to expand the students' vision. Improve students' learning autonomy and cultivate students' innovative thinking.
【学位授予单位】：云南师范大学
【学位级别】：硕士
【学位授予年份】：2013
【分类号】：G647

【参考文献】