[1]杨庚杭,沈苏彬.去除推荐场景多混淆因子的因果去偏方法[J].计算机技术与发展,2024,34(09):9-15.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0162]
　YANG Geng-hang,SHEN Su-bin.Causal Debias Method for Multiple Confounding Factors in Recommendation[J].,2024,34(09):9-15.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0162]
点击复制

去除推荐场景多混淆因子的因果去偏方法()

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

卷:
34
期数:
2024年09期
页码:
9-15
栏目:
大数据与云计算
出版日期:
2024-09-10

文章信息/Info

Title:
Causal Debias Method for Multiple Confounding Factors in Recommendation
文章编号:
1673-629X(2024)09-0009-07
作者:
杨庚杭1; 沈苏彬2
1. 南京邮电大学 计算机学院,江苏 南京 210023;2. 南京邮电大学 通信与网络技术国家工程研究中心,江苏 南京 210003
Author(s):
YANG Geng-hang1; SHEN Su-bin2
1. School of Computer Science,Nanjing University of Posts and Telecommunications,Nanjing 210023,China;2. National Engineering Research Center on Communication and Networking,Nanjing University of Posts and Telecommunications,Nanjing 210003,China
关键词:
因果推断; 前门调整; 后门调整; 多混淆因子; 因果图
Keywords:
causal inference; front-door adjustment; back-door adjustment; multiple confounding factors; causal graph
分类号:
TP391.3
DOI:
10.20165/j.cnki.ISSN1673-629X.2024.0162
摘要:
如何去除推荐场景中存在的偏差问题,是推荐系统提升效果的重大挑战。 现有的推荐模型只是拟合数据,基于相关性去除偏差会受到虚假相关性的影响;基于因果关系去除偏差则由于场景复杂很难抽取更加全面的因果关系。 因此,尽可能多地考虑各因素之间的因果关系并去除偏差问题很有必要。 该文从因果角度对用户行为在物品分类的分布不平衡和物品类别流行度两个混淆因子在推荐流程中的因果关系进行研究,提出考虑多因素的去混淆方法,有效去除推荐中的偏差问题。 首先,分析各变量之间的因果关系并构建因果图;其次,使用前门调整和后门调整去除两个混淆因子造成的虚假相关性以及偏差;最后,将该方法应用在神经网络因子分解机上,在两个公开数据集上进行了实验并验证。 从仿真实验结果可知,该方法相比于目前的最优方法都有不同程度的提升。
Abstract:
How to eliminate the bias problem existing in recommendation is a major challenge for recommender systems to improve their effectiveness. The existing recommendation models only fit the data, and eliminating bias based on correlation may be affected by spurious correlations. Eliminating bias based on causal relationships is difficult to extract more comprehensive causal relationships due to the complex scene. Therefore, it is necessary to consider the causal relationships between various factors as much as possible and eliminating bias issues. We study the causal relationship of the two confounding factors in the recommendation process,namely,the dis-tribution imbalance of user behavior in item classification and item category popularity. A de-confound method is proposed that considers multiple factors to effectively eliminate the bias problem in recommendation. Firstly,the causal relationship between variables in the rec-ommendation is analyzed and a causal graph is constructed. Secondly, the false correlation as well as the bias caused by the two confounding factors are eliminated using front-door adjustment and back-door adjustment. Finally,the proposed method is applied to neural factorization machines,and experimentally validated on two publicly available datasets, the results show that the proposed method has varying degrees of improvement compared to the current optimal method.

相似文献/References:

[1]李洪飞,万亚平,阳小华,等.一种基于 CDC 的适用于高维数据的因果推断算法[J].计算机技术与发展,2020,30(01):38.[doi:10. 3969 / j. issn. 1673-629X. 2020. 01. 007]
　LI Hong-fei,WAN Ya-ping,YANG Xiao-hua,et al.A High Dimensional Causal Inference Algorithm Based on CDC[J].,2020,30(09):38.[doi:10. 3969 / j. issn. 1673-629X. 2020. 01. 007]
[2]耿家兴,万亚平,李洪飞.基于神经网络的混合数据的因果发现[J].计算机技术与发展,2020,30(05):26.[doi:10. 3969 / j. issn. 1673-629X. 2020. 05. 006]
　GENG Jia-xing,WAN Ya-ping,LI Hong-fei.Causal Discovery of Mixed Data Based on Neural Network[J].,2020,30(09):26.[doi:10. 3969 / j. issn. 1673-629X. 2020. 05. 006]
[3]周 琦,万亚平,左建宏,等.基于因果推断肺癌患者预后治疗影响因素研究[J].计算机技术与发展,2021,31(08):145.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 025]
　ZHOU Qi,WAN Ya-ping,ZUO Jian-hong,et al.Research on Influencing Factors of Prognosis Treatment for Patients withLung Cancer Based on Causal Inference[J].,2021,31(09):145.[doi:10. 3969 / j. issn. 1673-629X. 2021. 08. 025]
[4]朱明朔,沈苏彬.一种基于因果推断的序列推荐模型[J].计算机技术与发展,2024,34(09):102.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0176]
　ZHU Ming-shuo,SHEN Su-bin.A Sequence Recommendation Model Based on Causal Inference[J].,2024,34(09):102.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0176]
[5]彭博,陆嘉伟,涂欣月,等.基于大数据分析的弓网参数反事实模型研究[J].计算机技术与发展,2025,(01):162.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0304]
　PENG Bo,LU Jia-wei,TU Xin-yue,et al.Research on Counterfactual Modeling of Pantograph Parameters Based on Big Data Analysis[J].,2025,(09):162.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0304]
[6]陈松,沈苏彬.一种改进强化学习算法的路径规划方法[J].计算机技术与发展,2025,(02):115.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0308]
　CHEN Song,SHEN Su-bin.A Path Planning Method Based on Improved Reinforcement Learning Algorithm[J].,2025,(09):115.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0308]
[7]郭亦嘉,沈苏彬.基于因果推断的弱监督野生动物识别算法研究[J].计算机技术与发展,2025,(05):158.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0394]
　GUO Yi-jia,SHEN Su-bin.Weakly Supervised Wildlife Recognition Algorithm Based on Causal Inference[J].,2025,(09):158.[doi:10.20165/j.cnki.ISSN1673-629X.2024.0394]

更新日期/Last Update: 2024-09-10