基于TF-IDF与用户聚类的推荐算法  

Recommendation Algorithm Based on TF-IDF and User Clustering

作　　者：林振荣[1] 黄虹霞 舒伟红 刘承启 LIN Zhen-rong;HUANG Hong-xia;SHU Wei-hong;LIU Cheng-qi(Information Engineering School,Nanchang University,Nanchang Jiangxi 330031,China)

机构地区：[1]南昌大学信息工程学院,江西南昌330031

出　　处：《计算机仿真》2022年第6期341-345,共5页Computer Simulation

摘　　要：针对基于用户协同过滤推荐算法未考虑物品特征对推荐效果存在影响的情况,提出基于TF-IDF(词频-逆文本频率指数)与用户聚类的推荐算法。利用TF-IDF算法得到物品的2类特征信息:用户-物品-特征TF值矩阵和特征的TF-IDF,将上述矩阵与用户身份属性信息合并后利用K-means聚类分析缩小用户集,并利用特征的TF-IDF值改进相似度计算公式,经计算后生成推荐列表。通过实验分析参数取不同值情况下对推荐算法效果的影响,并将该算法与传统的基于用户的协同过滤算法进行比较,能够验证所提出的推荐算法更优,最终结果表示上述算法可以得到不错的推荐效果。Aiming at the case where the recommendation algorithm based on user collaborative filtering does not consider the influence of item characteristics on the recommendation effect, a recommendation algorithm based on TF-IDF(term frequency-inverse document frequency) and user clustering is proposed. The TF-IDF algorithm was used to obtain 2 types of feature information of the article: user-item-feature TF value matrix and feature TF-IDF value. After combining the above matrix with user identity attribute information, K-means aggregation Class analysis was used to reduce the user set, and the TF-IDF value of the feature was used to improve the similarity calculation formula, and a recommendation list was generated after calculation. Through experiments, the effects of different values of parameters on the recommendation algorithm were analyzed, and the algorithm was compared with the traditional user-based collaborative filtering algorithm, which can verify that the proposed recommendation algorithm is better, and the final result shows that the algorithm can get a good recommended effect.

关 键 词：基于用户 物品特征信息 词频-逆文本频率指数 聚类 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]