Causal Embeddings for Recommendation

被引:170
|
作者
Bonner, Stephen [1 ,2 ]
Vasile, Flavian [1 ]
机构
[1] Criteo AI Labs, Paris, France
[2] Univ Durham, Dept Comp Sci, Durham, England
关键词
Recommender Systems; Causality; Embeddings; Neural Networks; Counterfactual Inference;
D O I
10.1145/3240323.3240360
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
Many current applications use recommendations in order to modify the natural user behavior, such as to increase the number of sales or the time spent on a website. This results in a gap between the final recommendation objective and the classical setup where recommendation candidates are evaluated by their coherence with past user behavior, by predicting either the missing entries in the user-item matrix, or the most likely next event. To bridge this gap, we optimize a recommendation policy for the task of increasing the desired outcome versus the organic user behavior. We show this is equivalent to learning to predict recommendation outcomes under a fully random recommendation policy. To this end, we propose a new domain adaptation algorithm that learns from logged data containing outcomes from a biased recommendation policy and predicts recommendation outcomes according to random exposure. We compare our method against state-of-the-art factorization methods, in addition to new approaches of causal recommendation and show significant improvements.
引用
收藏
页码:104 / 112
页数:9
相关论文
共 50 条
  • [1] Causal Embeddings for Recommendation: An Extended Abstract
    Vasile, Flavian
    Bonner, Stephen
    PROCEEDINGS OF THE TWENTY-EIGHTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2019, : 6236 - 6240
  • [2] Embeddings of Causal Sets
    Reid, David D.
    PROCEEDINGS OF THE NATIONAL SOCIETY OF BLACK PHYSICISTS, 2009, 1140 : 60 - 68
  • [3] On the Consistency of Average Embeddings for Item Recommendation
    Bendada, Walid
    Salha-Galvan, Guillaume
    Hennequin, Romain
    Bouabca, Thomas
    Cazenave, Tristan
    PROCEEDINGS OF THE 17TH ACM CONFERENCE ON RECOMMENDER SYSTEMS, RECSYS 2023, 2023, : 833 - 839
  • [4] Are GPT Embeddings Useful for Ads and Recommendation?
    Peng, Wenjun
    Xu, Derong
    Xu, Tong
    Zhang, Jianjin
    Chen, Enhong
    KNOWLEDGE SCIENCE, ENGINEERING AND MANAGEMENT, PT IV, KSEM 2023, 2023, 14120 : 151 - 162
  • [5] Location Embeddings for Next Trip Recommendation
    Dadoun, Amine
    Troncy, Raphael
    Ratier, Olivier
    Petitti, Riccardo
    COMPANION OF THE WORLD WIDE WEB CONFERENCE (WWW 2019 ), 2019, : 896 - 903
  • [6] Adapting Text Embeddings for Causal Inference
    Veitch, Victor
    Sridhar, Dhanya
    Blei, David M.
    CONFERENCE ON UNCERTAINTY IN ARTIFICIAL INTELLIGENCE (UAI 2020), 2020, 124 : 919 - 928
  • [7] A survey on causal inference for recommendation
    Luo, Huishi
    Zhuang, Fuzhen
    Xie, Ruobing
    Zhu, Hengshu
    Wang, Deqing
    An, Zhulin
    Xu, Yongjun
    INNOVATION, 2024, 5 (02):
  • [8] Convolutional Gaussian Embeddings for Personalized Recommendation with Uncertainty
    Jiang, Junyang
    Yang, Deqing
    Xiao, Yanghua
    Shen, Chenlu
    PROCEEDINGS OF THE TWENTY-EIGHTH INTERNATIONAL JOINT CONFERENCE ON ARTIFICIAL INTELLIGENCE, 2019, : 2642 - 2648
  • [9] Job Recommendation Based on Extracted Skill Embeddings
    Kara, Atakan
    Danis, F. Serhan
    Orman, Gunce K.
    Turhan, Sultan N.
    Ozlu, O. Anil
    INTELLIGENT SYSTEMS AND APPLICATIONS, VOL 3, 2023, 544 : 497 - 507
  • [10] Using Tweets Embeddings For Hashtag Recommendation in Twitter
    Ben-Lhachemi, Nada
    Nfaoui, El Habib
    PROCEEDINGS OF THE FIRST INTERNATIONAL CONFERENCE ON INTELLIGENT COMPUTING IN DATA SCIENCES (ICDS2017), 2018, 127 : 7 - 15