基于时间感知和社交网络信任度的协同过滤算法研究

发布时间：2018-01-22 03:35

本文关键词： 推荐系统协同过滤算法时间感知社交网络　出处：《南京信息工程大学》2016年硕士论文　论文类型：学位论文

【摘要】：随着互联网的迅速发展,推荐系统已经广泛地应用在电影、音乐以及电子商务网站等众多领域中。协同过滤算法是推荐系统中最为流行的技术,在互联网的众多领域发挥着核心作用。它能够根据用户的历史评分数据搜索相似的用户或物品,然后进行评分预测。尽管协同过滤算法已被成功地应用到众多的电子商务推荐系统中,但仍然面临着诸如数据稀疏性、冷启动以及用户兴趣变化等问题。如何有效的解决这些问题已经成为一个重要的研究课题。同时,目前各种社交网络平台纷纷涌现并日益渗透到人们的日常生活中,尤其影响着人们在电子商务系统上选择和购买物品的行为。从社交网络这个全新的角度来重新考察现有的产品推荐系统是非常有意义的。针对以上重要问题,本文主要进行了以下几个方面的研究。第一,基于时间感知的协同过滤算法研究。现如今,推荐系统呈现动态化趋势,用户偏好随时间发生改变,所以本文将时间信息融合到协同过滤算法中。同时,传统的协同过滤算法在计算用户间相似度时没有考虑到不同目标物品的影响,导致一个用户对于所有的目标物品的近邻用户是完全相同的。本文在解决一个目标用户对于不同的目标物品应该具有不同的近邻这一问题的同时结合时间感知提出了一个新的相似度计算公式。通过一系列的实验证明,改进的算法在精度、召回率、MAE以及覆盖度这几个方面优于传统的协同过滤算法。第二,基于社交网络信任度的协同过滤算法研究。针对传统协同过滤算法面临的冷启动问题,提出一种新的融合社交网路信任度的协同过滤算法。首先利用用户社交网络信息建模,然后结合用户信任度以及偏好来进行评分预测。最后通过两种不同方式的实验验证,相比于传统的协同过滤算法,该算法不仅提高了推荐的准确度,还能够有效地缓解数据稀疏性以及冷启动问题。
[Abstract]:With the rapid development of Internet, recommendation system has been widely used in many fields, such as movies, music and e-commerce websites. Collaborative filtering algorithm is the most popular technology in recommendation system. It plays a central role in many areas of the Internet. It can search for similar users or items based on users' historical scoring data. Although collaborative filtering algorithm has been successfully applied to many e-commerce recommendation systems, it still faces data sparsity. How to solve these problems effectively has become an important research topic. At the same time. At present, a variety of social networking platforms have emerged and increasingly penetrated into people's daily life. In particular, it affects the behavior of people to choose and purchase items in electronic commerce system. It is very meaningful to re-examine the existing product recommendation system from the new perspective of social network. This paper mainly studies the following aspects. First, the collaborative filtering algorithm based on time perception. Nowadays, recommendation system presents a dynamic trend, user preferences change with time. Therefore, the time information is fused into the collaborative filtering algorithm. At the same time, the traditional collaborative filtering algorithm does not take into account the influence of different target items when calculating the similarity between users. This paper solves the problem that a target user should have different nearest neighbors for different target items at the same time combined with time perception. A new formula for calculating similarity is given, which is proved by a series of experiments. The improved algorithm is superior to the traditional collaborative filtering algorithm in precision, recall rate and coverage. The research of collaborative filtering algorithm based on trust degree of social network. Aiming at the cold start problem of traditional collaborative filtering algorithm. In this paper, a new collaborative filtering algorithm based on trust degree of social network is proposed. Firstly, user social network information is used to model the model. Finally, through two different ways of experimental verification, compared with the traditional collaborative filtering algorithm, this algorithm not only improves the accuracy of recommendation. It can also effectively alleviate the data sparsity and cold start problem.
【学位授予单位】：南京信息工程大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP391.3

【相似文献】