基于多特征排序模型的网络课程推荐算法研究与应用

发布时间：2019-01-23 08:31

【摘要】：随着大规模网络开放课程的飞速发展,在线教育这一全新的学习形式开始被越来越多的人所接受。用户通过互联网可以学习到各种领域的知识和技能,但随着在线课程资源数量的增加以及种类的越来越多样化,用户在考虑想要学习的课程时经常会遇到选择难题。推荐算法的引入能够为用户的课程学习提供建议,但由于网络课程存在一些局限性,如文本信息较少、用户行为信息不够丰富、评价信息缺乏等,传统的推荐算法无法直接应用到网络课程的推荐中,需要基于网络课程的独特场景进行创新和改进。本文在对云课堂用户数据进行了充分分析的基础上,研究并实现了一种基于多特征排序模型的网络课程推荐算法。该算法结合了网络课程及用户相关的多个特征,包括基于主题的用户偏好、基于协同过滤的用户偏好、课程热门度、讲师影响力。通过排序学习的方法对这些特征进行线性组合,计算目标用户与网络课程间的匹配程度,从而为用户进行课程推荐。为了验证算法的有效性,本文在云课堂真实数据集上进行了大量实验,实验证明本文算法能够得到较好的推荐效果,与参照算法相比有一定提升。另外,本文设计实现了基于云课堂的课程推荐系统,其功能是在云课堂用户个人学习主页的基础上实现的。经测试系统运行良好,验证了本文算法的实用性。
[Abstract]:With the rapid development of open online courses, online education, a new form of learning, has been accepted by more and more people. Users can learn knowledge and skills in various fields through the Internet, but with the increasing number and variety of online course resources, users often encounter difficulties in choosing the courses they want to learn. The introduction of recommendation algorithm can provide suggestions for users to learn courses. However, there are some limitations in web-based courses, such as less text information, insufficient information on user behavior, lack of evaluation information, etc. The traditional recommendation algorithm can not be directly applied to the recommendation of network courses, and it needs to be innovated and improved based on the unique scene of network courses. Based on the full analysis of cloud classroom user data, this paper studies and implements a network course recommendation algorithm based on multi-feature ranking model. The algorithm combines several features of network courses and users, including topic-based user preferences, collaborative filtering based user preferences, course popularity, instructor influence. The linear combination of these features is carried out by the method of ranking learning, and the matching degree between the target user and the network course is calculated, and then the course recommendation is made for the user. In order to verify the effectiveness of the algorithm, a large number of experiments have been carried out on the real data set in the cloud classroom. The experimental results show that the proposed algorithm can get a better recommendation effect and has a certain improvement compared with the reference algorithm. In addition, this paper designs and implements the course recommendation system based on cloud classroom, whose function is realized on the basis of the home page of cloud classroom user's personal learning. The test system runs well, and the practicability of the algorithm is verified.
【学位授予单位】：浙江大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【相似文献】