Publication Type

Journal Article

Version

publishedVersion

Publication Date

1-2024

Abstract

Recommender systems easily face the issue of user preference shifts. User representations will become outof-date and lead to inappropriate recommendations if user preference has shifted over time. To solve theissue, existing work focuses on learning robust representations or predicting the shifting pattern. Therelacks a comprehensive view to discover the underlying reasons for user preference shifts. To understand thepreference shift, we abstract a causal graph to describe the generation procedure of user interaction sequences.Assuming user preference is stable within a short period, we abstract the interaction sequence as a set ofchronological environments. From the causal graph, we find that the changes of some unobserved factors (e.g.,becoming pregnant) cause preference shifts between environments. Besides, the fine-grained user preferenceover item categories sparsely affects the interactions with different items. Inspired by the causal graph,our key considerations to handle preference shifts lie in modeling the interaction generation procedure by:(1) capturing the preference shifts across environments for accurate preference prediction and(2) disentangling the sparse influence from user preference to interactions for accurate effect estimationof preference. To this end, we propose a Causal Disentangled Recommendation (CDR) framework, whichcaptures preference shifts via a temporal variational autoencoder and learns the sparse influence frommultiple environments. Specifically, an encoder is adopted to infer the unobserved factors from userinteractions while a decoder is to model the interaction generation process. Besides, we introduce twolearnable matrices to disentangle the sparse influence from user preference to interactions. Last, we devise amulti-objective loss to optimize CDR. Extensive experiments on three datasets show the superiority of CDRin enhancing the generalization ability under user preference shifts.

Keywords

Causal disentangled recommendation, preference shifts, generalizable recommendation, out-of-distribution generalization

Discipline

Artificial Intelligence and Robotics | Databases and Information Systems

Research Areas

Intelligent Systems and Optimization

Areas of Excellence

Digital transformation

Publication

ACM Transactions on Information Systems

Volume

42

Issue

1

First Page

1

Last Page

27

ISSN

1046-8188

Identifier

10.1145/3593022

Publisher

Association for Computing Machinery (ACM)

Additional URL

https://doi.org/10.1145/3593022

Download

Share

COinS