一种基于上下文的推荐算法的研究

发布时间：2018-09-05 14:04

【摘要】：近年来,随着互联网的发展,网上购物日渐成为人们生活中必不可少的存在,随着电子商务网站中用户和物品数目的大量增长,如何迅速推荐用户感兴趣的物品已成为急需解决的问题。传统推荐算法更多的关注用户评分,对用户的上下文信息利用不足,本课题在深入研究已有个性化推荐技术的基础上,针对用户的上下文情景,研究并设计了一种基于上下文的协同过滤推荐算法。首先,本文阐述了个性化推荐系统中常用的各种推荐技术尤其是协同过滤技术,充分了解了协同过滤算法的原理及分类。详细介绍了几种经典算法的具体过程,分析了这些算法的优缺点。然后介绍了上下文感知推荐系统的相关技术,最后介绍了常用的推荐算法质量评估方法。其次,本文提出了一种含有上下文权重的上下文相似度计算方法,针对现有上下文引入方法的不足,给出了一种上下文预过滤与上下文建模相结合的上下文引入方法,并提出了一种基于上下文相似度的Slope One改进算法,进行数据预测填充,降低数据稀疏性,以降低数据稀疏性对算法的不利影响。再次,通过对经典协同过滤推荐算法的研究,发现传统的协同过滤算法中存在的不足,结合上下文相似度,提出一种基于上下文相似度的协同过滤推荐算法。相对于传统的推荐系统,利用上下文相似度挖掘用户兴趣偏好更加准确,在很大程度上提高了推荐精度。最后,通过实验对改进模型及算法的性能进行验证,并与其它算法进行对比,分析评估结果,验证了算法在一定程度上得到较好的推荐预测效果。同时也检验了算法在不同数据集上的性能,验证了算法的健壮性,本课题研究可提升协同过滤推荐算法的效率,对个性化推荐技术的实践应用提供了一定的理论及方法支持。
[Abstract]:In recent years, with the development of the Internet, online shopping has become an indispensable existence in people's lives, and with the increase of the number of users and items in e-commerce websites, How to quickly recommend items of interest to users has become an urgent problem. The traditional recommendation algorithm pays more attention to the user's score and makes insufficient use of the user's context information. Based on the in-depth study of the existing personalized recommendation technology, this paper aims at the user's context situation. A context-based collaborative filtering recommendation algorithm is studied and designed. Firstly, this paper describes various recommendation technologies, especially collaborative filtering technologies, which are commonly used in personalized recommendation systems, and fully understands the principle and classification of collaborative filtering algorithms. The process of several classical algorithms is introduced in detail, and the advantages and disadvantages of these algorithms are analyzed. Then it introduces the relevant technology of context-aware recommendation system, and finally introduces the commonly used quality evaluation methods of recommendation algorithms. Secondly, a context similarity calculation method with context weight is proposed. Aiming at the shortcomings of existing context introduction methods, a context introduction method combining context prefiltering and context modeling is proposed. An improved Slope One algorithm based on context similarity is proposed to reduce the data sparsity and reduce the data sparsity in order to reduce the adverse effect of data sparsity on the algorithm. Thirdly, through the research of the classical collaborative filtering recommendation algorithm, the shortcomings of the traditional collaborative filtering algorithm are found. A collaborative filtering recommendation algorithm based on the context similarity is proposed. Compared with the traditional recommendation system, it is more accurate to mine user interest preference by using context similarity, which improves the recommendation accuracy to a great extent. Finally, the performance of the improved model and algorithm is verified by experiments, and compared with other algorithms, the evaluation results are analyzed and the results show that the proposed algorithm has a better prediction effect to a certain extent. At the same time, the performance of the algorithm on different data sets is tested, and the robustness of the algorithm is verified. This research can improve the efficiency of collaborative filtering recommendation algorithm, and provide some theoretical and methodological support for the application of personalized recommendation technology.
【学位授予单位】：燕山大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【相似文献】