[1]贾俊康,李玲娟.结合贡献度与时间权重的协同过滤推荐算法[J].计算机技术与发展,2023,33(03):167-172.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 025]
　JIA Jun-kang,LI Ling-juan.Collaborative Filtering Recommendation Algorithm Combining Contribution and Time Weight[J].,2023,33(03):167-172.[doi:10. 3969 / j. issn. 1673-629X. 2023. 03. 025]
点击复制

结合贡献度与时间权重的协同过滤推荐算法()
分享到：

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

卷:
33
期数:
2023年03期
页码:
167-172
栏目:
人工智能
出版日期:
2023-03-10

文章信息/Info

Title:
Collaborative Filtering Recommendation Algorithm Combining Contribution and Time Weight
文章编号:
1673-629X(2023)03-0167-06
作者:
贾俊康; 李玲娟
南京邮电大学 计算机学院,江苏 南京 210023
Author(s):
JIA Jun-kang; LI Ling-juan
School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China
关键词:
贡献度; 时间因子; 相似度; 协同过滤; 推荐
Keywords:
contribution degree; time factor; similarity; collaborative filtering; recommendation
分类号:
TP311
DOI:
10. 3969 / j. issn. 1673-629X. 2023. 03. 025
摘要:
传统的协同过滤推荐算法未考虑用户兴趣随时间动态变化,以及不同用户对同一项目评分差异过大对推荐效果的影响,导致推荐效果不理想。 针对以上问题,以进一步提高基于用户的协同过滤推荐算法的精度为目标,设计了一种结合贡献度与时间权重的协同过滤推荐算法 CTCF。 该算法在用户相似度计算中引入可信误差阈值、贡献度与时间权重。首先,利用用户评分信息构建用户-评分矩阵与用户-评分时间矩阵,依据可信误差阈值来计算用户贡献度;然后,引入拟合贡献度与时间因子的遗忘曲线得到时间权重,再将时间权重引入皮尔逊相关系数中计算用户相似度;找出目标用户的邻居集,并预测目标用户对邻居集对应项目中未评分项目的评分;最后,按评分由高到低生成 Top-N 推荐。 在 MovieLens数据集上的测试结果表明,CTCF 算法具有更高的 F1 值,有效提高了推荐精度和动态性
Abstract:
The traditional collaborative filtering recommendation algorithm does not consider the dynamic changes of users’ interests overtime and the impact of excessive differences in the scores of different users on the same item on the recommendation effect,resulting inthe unsatisfactory recommendation effect. In view of the above problems, a collaborative filtering recommendation algorithm CTCF,which combines contribution degree and time weight,is designed to further improve the accuracy 　of user-based collaborative filtering recommendation algorithm. The algorithm introduces the trusted error threshold,contribution degree and time weight into the user similaritycalculation. Firstly,the user score matrix and user score time matrix are constructed by using the user score information,and the user contribution degree is calculated by the trusted error threshold. Then,the forgetting curve of fitting contribution degree and time factor is introduced to obtain the time weight, and the time weight is introduced into the Pearson correlation coefficient to calculate the usersimilarity,find out the neighbor set of the target user,and predict the score of the target user on the non-scored items in the correspondingitems of the neighbor set. Finally,the Top-N recommendation is generated according to the score from high to low. 　The test results onMovieLens dataset show that CTCF algorithm has higher F1 value and effectively improves the recommendation precision and dynamics.

相似文献/References:

[1]付常雷.一种基于 Newman 快速算法改进的社团划分算法[J].计算机技术与发展,2018,28(01):33.[doi:10.3969/ j. issn.1673-629X.2018.01.007]
　FU Chang-lei.A Community Partitioning Algorithm Based on Improved Fast-Newman Algorithm[J].,2018,28(03):33.[doi:10.3969/ j. issn.1673-629X.2018.01.007]
[2]刘金梅,舒远仲,张尚田.基于评分填充和时间的加权 Slope One 算法[J].计算机技术与发展,2021,31(01):35.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 007]
　LIU Jin-mei,SHU Yuan-zhong,ZHANG Shang-tian.A Weighted Slope One Algorithm Based on Rating Filling and Time[J].,2021,31(03):35.[doi:10. 3969 / j. issn. 1673-629X. 2021. 01. 007]
[3]李 浩,梁京章,潘 莹 *.一种改进的兴趣相似度个性化推荐算法[J].计算机技术与发展,2022,32(12):1.[doi:10. 3969 / j. issn. 1673-629X. 2022. 12. 001]
　LI Hao,LIANG Jing-zhang,PAN Ying *.An Improved Personalized Recommendation Algorithm Based on Interest Similarity[J].,2022,32(03):1.[doi:10. 3969 / j. issn. 1673-629X. 2022. 12. 001]

更新日期/Last Update: 2023-03-10