基于隐性社交圈的推荐算法

Recommendation Algorithm Based on Implicit Social Circle

用户之间的社交信息已经被广泛地应用于改进传统的推荐系统。然而,在许多网站,如亚马逊和eBay,没有明确的社交图表可以用来提高推荐性能。因此,在这项工作中,为了使非社交性网站能够采用社交推荐的方法,论文提出了一个通用的框架,根据用户对所购商品给出的评分和评论来构建一个隐性社交圈。并将这种隐性的社交圈融合到目前已有的显性社交推荐算法中,从而来增强任何没有社交网络的推荐系统的性能。该算法通过皮尔逊相关系数PCC(Pearson Correla⁃tion Coefficient)来分析和提取用户之间的隐性社交圈,并将其融合于基于矩阵分解的社交推荐RS(Social Regularization)算法中,然后在Amazon数据集上进行实验。实验结果表明,该方法比传统的无社会信息推荐方法更有效。

Social information between users has been widely used to improve traditional recommendation systems.However,on many websites,such as Amazon and eBay,there are no clear social graphs that can be used to improve recommendation perfor⁃mance.Therefore,in this work,in order to enable non-social websites to adopt social recommendation methods,a general frame⁃work is proposed to construct a hidden social circle based on the users'ratings and comments on the purchased products.This implic⁃it social circle is incorporated into the existing explicit social recommendation algorithms to enhance the performance of any recom⁃mendation system without social networks.The algorithm analyzes and extracts the implicit social circle between users by Pearson Correlation Coefficient(PCC)and fuses it into the social decomposition RS(Social Regularization)algorithm based on matrix de⁃composition,and then the experiment is done on the Amazon data set.The experimental results show that this method is more effec⁃tive than the traditional no-social information recommendation method.