移动电子商务环境下基于数据分析的商品推荐算法

发布时间：2018-05-05 05:27

本文选题：协同过滤 + 推荐算法　；参考：《北京邮电大学》2017年硕士论文

【摘要】：随着互联网的不断发展和人们消费观念的转变,电子商务平台成为了人们首要的购物方式。在这种情况下,个性化推荐系统成为了帮助用户快速发现喜好商品的主要方式,同时也成为了帮助平台销售商品的主要手段。在推荐系统中,协同过滤推荐算法是被最广泛使用的,该算法的主要思想是寻找目标用户的相似邻居集,将该邻居集的用户喜好作为推荐项推荐给目标用户。但是,随着移动互联网的高速发展,传统的协同过滤推荐算法面临着更多挑战。首先,在移动互联网环境下,电子商务平台可以获取更多关于用户的信息内容,诸如用户位置上下文信息。这些信息对推荐算法中计算用户之间的相似性产生了巨大的影响。其次,移动互联网为推荐算法带来了更加庞大的数据量,在大数据环境下,推荐算法的推荐实时性难以得到有效的保障。针对上述移动互联网环境下推荐算法面临的两个问题,本文进行了深入的研究分析,提出了相应的解决方案。针对移动互联网环境下用户位置上下文信息对推荐算法中用户间相似性计算影响的问题,本文提出一种基于位置上下文信息的协同过滤推荐算法(CFLC, A Collaborative Filtering Algorithm Based on Location Context)。该算法将用户位置信息作为关键影响因素,与修正的余弦相似性计算方法结合,提出基于位置上下文信息的相似性计算方法,并利用这种相似性计算方法计算用户之间的相似性,进而寻找到目标用户的最近邻居集,最后通过最近邻居集中用户的兴趣爱好项为目标用户提供推荐。针对移动互联网环境下电子商务平台产生庞大数据量的问题,对传统协同过滤推荐算法造成严重的负载影响,本文提出了一种基于并行化迭代式k-medoids聚类的协同过滤推荐算法(CFPKM,A Collaborative Filtering Algorithm Based on Parallelized k-medoids Clustering)。该算法使用改进的并行化k-medoids聚类算法,预先对推荐算法中的用户-项目评分矩阵进行聚类分析,将用户分为k类。然后,计算目标用户与所属类中的其他用户之间的相似性,通过缩小数据空间来提高整个推荐算法的准确度和推荐速度。最后,本文通过对提出算法的实验仿真,以及与传统的协同过滤推荐算法进行比较分析,证明了在考虑用户位置信息以及利用聚类方式预先处理数据的情况下,进一步提高了推荐算法的有效性和准确性。
[Abstract]:With the continuous development of the Internet and the change of people's consumption concept, e-commerce platform has become the most important way of shopping. In this case, personalized recommendation system has become the main way to help users quickly find favorite products, but also become the main means to help the platform to sell goods. In the recommendation system, collaborative filtering recommendation algorithm is the most widely used. The main idea of the algorithm is to find the similar neighbor set of the target user, and recommend the user preference of the neighbor set to the target user as a recommendation item. However, with the rapid development of mobile Internet, the traditional collaborative filtering recommendation algorithm is facing more challenges. First, in the mobile Internet environment, e-commerce platform can obtain more information about users, such as user location context information. This information has a great impact on computing the similarity between users in the recommendation algorithm. Secondly, the mobile Internet brings a larger amount of data to the recommendation algorithm. Under big data environment, the real-time recommendation of the recommendation algorithm is difficult to be effectively guaranteed. In view of the two problems faced by the recommendation algorithm in the mobile Internet environment, this paper makes a thorough study and analysis, and puts forward the corresponding solution. In order to solve the problem of the influence of user location context information on the similarity calculation between users in the recommendation algorithm, this paper proposes a collaborative filtering recommendation algorithm based on location context information (CFLC, A Collaborative Filtering Algorithm Based on Location context). The algorithm combines the user location information with the modified cosine similarity calculation method, and proposes a similarity calculation method based on location context information. The similarity calculation method is used to calculate the similarity between the users, and then the nearest neighbor set of the target user is found. Finally, the user's interests and hobbies in the nearest neighbor set are recommended to the target user. In order to solve the problem of large amount of data generated by e-commerce platform in the mobile Internet environment, the traditional collaborative filtering recommendation algorithm is seriously affected by the load. In this paper, a collaborative filtering recommendation algorithm based on parallel iterative k-medoids clustering is proposed. A Collaborative Filtering Algorithm Based on Parallelized k-medoids clustering is proposed. The improved parallelized k-medoids clustering algorithm is used to cluster the user-item scoring matrix in the recommendation algorithm and the users are divided into k-class. Then, the similarity between the target user and other users in the class is calculated, and the accuracy and speed of the entire recommendation algorithm are improved by reducing the data space. Finally, through the experimental simulation of the proposed algorithm and the comparative analysis with the traditional collaborative filtering recommendation algorithm, it is proved that in the case of considering the user location information and using clustering to pre-process the data, The effectiveness and accuracy of the recommendation algorithm are further improved.
【学位授予单位】：北京邮电大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP391.3

【相似文献】