TY - GEN
T1 - A modular adversarial approach to social recommendation
AU - Krishnan, Adit
AU - Cheruvu, Hari
AU - Tao, Cheng
AU - Sundaram, Hari
N1 - Publisher Copyright:
© 2019 Association for Computing Machinery.
PY - 2019/11/3
Y1 - 2019/11/3
N2 - This paper proposes a novel framework to incorporate social regularization for item recommendation. Social regularization grounded in ideas of homophily and influence appears to capture latent user preferences. However, there are two key challenges: first, the importance of a specific social link depends on the context and second, a fundamental result states that we cannot disentangle homophily and influence from observational data to determine the effect of social inference. Thus we view the attribution problem as inherently adversarial where we examine two competing hypothesis-social influence and latent interests-to explain each purchase decision. We make two contributions. First, we propose a modular, adversarial framework that decouples the architectural choices for the recommender and social representation models, for social regularization. Second, we overcome degenerate solutions through an intuitive contextual weighting strategy, that supports an expressive attribution, to ensure informative social associations play a larger role in regularizing the learned user interest space. Our results indicate significant gains (5-10% relative Recall@K) over state-of-the-art baselines across multiple publicly available datasets.
AB - This paper proposes a novel framework to incorporate social regularization for item recommendation. Social regularization grounded in ideas of homophily and influence appears to capture latent user preferences. However, there are two key challenges: first, the importance of a specific social link depends on the context and second, a fundamental result states that we cannot disentangle homophily and influence from observational data to determine the effect of social inference. Thus we view the attribution problem as inherently adversarial where we examine two competing hypothesis-social influence and latent interests-to explain each purchase decision. We make two contributions. First, we propose a modular, adversarial framework that decouples the architectural choices for the recommender and social representation models, for social regularization. Second, we overcome degenerate solutions through an intuitive contextual weighting strategy, that supports an expressive attribution, to ensure informative social associations play a larger role in regularizing the learned user interest space. Our results indicate significant gains (5-10% relative Recall@K) over state-of-the-art baselines across multiple publicly available datasets.
KW - Adversarial Machine Learning
KW - Generative Adversarial Networks
KW - Neural Collaborative Filtering
KW - Social Recommendation
UR - http://www.scopus.com/inward/record.url?scp=85075427548&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=85075427548&partnerID=8YFLogxK
U2 - 10.1145/3357384.3357898
DO - 10.1145/3357384.3357898
M3 - Conference contribution
AN - SCOPUS:85075427548
T3 - International Conference on Information and Knowledge Management, Proceedings
SP - 1753
EP - 1762
BT - CIKM 2019 - Proceedings of the 28th ACM International Conference on Information and Knowledge Management
PB - Association for Computing Machinery
T2 - 28th ACM International Conference on Information and Knowledge Management, CIKM 2019
Y2 - 3 November 2019 through 7 November 2019
ER -