[1]唐 震,黄 刚,华雯丽.融合协同过滤的 CatBoost 推荐算法[J].计算机技术与发展,2021,31(09):36-42.[doi:10. 3969 / j. issn. 1673-629X. 2021. 09. 007]
　TANG Zhen,HUANG Gang,HUA Wen-li.CatBoost Recommendation Algorithm with Collaborative Filtering[J].,2021,31(09):36-42.[doi:10. 3969 / j. issn. 1673-629X. 2021. 09. 007]
点击复制

融合协同过滤的 CatBoost 推荐算法()

《计算机技术与发展》[ISSN:1006-6977/CN:61-1281/TN]

卷:
31
期数:
2021年09期
页码:
36-42
栏目:
大数据分析与挖掘
出版日期:
2021-09-10

文章信息/Info

Title:
CatBoost Recommendation Algorithm with Collaborative Filtering
文章编号:
1673-629X(2021)09-0036-07
作者:
唐 震; 黄 刚; 华雯丽
南京邮电大学 计算机学院、软件学院、网络空间安全学院,江苏 南京 210023
Author(s):
TANG Zhen; HUANG Gang; HUA Wen-li
School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China
关键词:
CatBoost; 协同过滤; 准确性; 推荐系统; 混合模型
Keywords:
CatBoost; collaborative filtering; accuracy; recommendation system; hybrid model
分类号:
TP309
DOI:
10. 3969 / j. issn. 1673-629X. 2021. 09. 007
摘要:
在推荐系统中,针对推荐准确度问题,提出了一种融合协同过滤和 CatBoost 的混合推荐算法( UCF-CB) 。 在协同过滤模块中对用户相似度计算公式进行改进,加入时间衰减因子以及热门物品惩罚项,利用改进后的协同过滤算法对用户项目评分矩阵进行评分预测,得到用户对物品的一次评分。 对协同过滤一次评分进行降序排序,选取评分最高的前 k 项物品,形成召回集。 对原始数据集进行预处理,挖掘潜在特征增加特征维度,利用 CatBoost 算法对用户和项目特征进行训练,对召回集数据进行预测,得到二次评分预测。 对于没有评分记录的新用户,利用训练好的 CatBoost 算法可以直接进行评分预测,在一定程度上解决了推荐系统冷启动的问题。 将协同过滤一次评分以及 CatBoost 二次预测评分进行加权融合得到更为精确的推荐结果。 在 movieslens( ml-1m) 数据集上的实验结果表明,该算法可以获得较高的准确度。
Abstract:
In the recommendation system,a hybrid recommendation algorithm (UCF-CB) which combines collaborative filtering and CatBoost is proposed to improve the accuracy of recommendation algorithm. In the collaborative filtering module,the user similarity calculation formula is improved by adding time attenuation factor and hot item penalty factor. The improved collaborative filtering algorithm is used to predict the user’s item rating matrix,then the user’s first prediction score is obtained. The first score of collaborative filtering is sorted in descending order,and the top k items with the highest score are selected to form the recall set. After preprocessing the original data set,mining potential features and increasing feature dimensions,CatBoost algorithm is used to train the features of users and items,the trained model is used to predict the recall set data,and the second score prediction is obtained. For the new users without scoring records,the trained CatBoost algorithm can directly predict the score,which solves the problem of cold start of recommendation system to some extent. The first prediction score of collaborative filtering and the second prediction score of CatBoost are weighted to get more accurate recommendation results. Experiments on movieslens ( ml-1m) data set show that the proposed algorithm can achieve high accuracy.

相似文献/References:

[1]邵延振 蒙韧 袁鼎荣 李新友.基于Web结构分区的协同过滤推荐算法研究[J].计算机技术与发展,2010,(06):67.
　SHAO Yan-zhen,MENG Ren,YUAN Ding-rong,et al.Collaborative Filtering Recommendation Algorithm Research Based on Web Blocks[J].,2010,(09):67.
[2]查文琴 梁昌勇 曹镭.基于用户聚类的协同过滤推荐方法[J].计算机技术与发展,2009,(06):69.
　ZHA Wen-qin,LIANG Chang-yong,CAO Lei.Collaborative Filtering Recommendation Method Based on Clustering of Users[J].,2009,(09):69.
[3]姜雅倩 王直杰 张珏.基于供求关系及协同过滤技术的推荐模型研究[J].计算机技术与发展,2007,(06):18.
　JIANG Ya-qian,WANG Zhi-jie,ZHANG Jue.Research on Recommendation Model Based on Supply and Demand Relation and Collaborative Filtering[J].,2007,(09):18.
[4]游文 叶水生.电子商务推荐系统中的协同过滤推荐[J].计算机技术与发展,2006,(09):70.
　YOU Wen,YE Shui-sheng.A Survey of Collaborative Filtering Algorithm Applied in E- commerce Recommender System[J].,2006,(09):70.
[5]徐红 彭黎 郭艾寅 徐云剑.基于用户多兴趣的协同过滤策略改进研究[J].计算机技术与发展,2011,(04):73.
　XU Hong,PENG Li,GUO Ai-yin,et al.User-Based Collaborative Filtering Strategies More Interested in Improvement of Research[J].,2011,(09):73.
[6]杨东风 牛永洁.基于混合规则的图书推荐模型设计与研究[J].计算机技术与发展,2011,(07):210.
　YANG Dong-feng,NIU Yong-jie.Books Recommended Model Design and Research Based on Mixing Rules[J].,2011,(09):210.
[7]吴月萍 王娜 马良.基于蚁群算法的协同过滤推荐系统的研究[J].计算机技术与发展,2011,(10):73.
　WU Yue-ping,WANG Na,MA Liang.Research of Collaboration Filtering Recommendation System Based on Ant Algorithm[J].,2011,(09):73.
[8]李克潮,蓝冬梅.一种属性和评分的协同过滤混合推荐算法[J].计算机技术与发展,2013,(07):116.
　LI Ke-chao,LAN Dong-mei.A Collaborative Filtering Hybrid Recommendation Algorithm for Attribute and Rating[J].,2013,(09):116.
[9]范虎,花伟伟.协同过滤推荐算法的研究与改进[J].计算机技术与发展,2013,(09):66.
　FAN Hu[],HUA Wei-wei[].Research and Improvement of Collaborative Filtering Recommendation Algorithm[J].,2013,(09):66.
[10]李振博,徐桂琼,査九. 基于用户谱聚类的协同过滤推荐算法[J].计算机技术与发展,2014,24(09):59.
　LI Zhen-bo,XU Gui-qiong,ZHA Jiu. A Collaborative Filtering Recommendation Algorithm Based on User Spectral Clustering[J].,2014,24(09):59.

更新日期/Last Update: 2021-09-10