Disentangled Interest Network for Out-of-Distribution CTR Prediction

TL;DR

This paper introduces DiseCTR, a disentangled interest network that models multiple user interests to improve out-of-distribution CTR prediction accuracy and robustness using a causal framework.

Contribution

It proposes a novel causal disentanglement approach for CTR prediction that effectively handles evolving user interests and out-of-distribution data.

Findings

Achieves over 0.02 higher AUC and GAUC than baselines.

Reduces logloss by over 13.7% on real datasets.

Successfully disentangles user interests for better OOD generalization.

Abstract

Click-through rate (CTR) prediction, which estimates the probability of a user clicking on a given item, is a critical task for online information services. Existing approaches often make strong assumptions that training and test data come from the same distribution. However, the data distribution varies since user interests are constantly evolving, resulting in the out-of-distribution (OOD) issue. In addition, users tend to have multiple interests, some of which evolve faster than others. Towards this end, we propose Disentangled Click-Through Rate prediction (DiseCTR), which introduces a causal perspective of recommendation and disentangles multiple aspects of user interests to alleviate the OOD issue in recommendation. We conduct a causal factorization of CTR prediction involving user interest, exposure model, and click model, based on which we develop a deep learning implementation for these three causal mechanisms. Specifically, we first design an interest encoder with sparse attention which maps raw features to user interests, and then introduce a weakly supervised interest disentangler to learn independent interest embeddings, which are further integrated by an attentive interest aggregator for prediction. Experimental results on three real-world datasets show that DiseCTR achieves the best accuracy and robustness in OOD recommendation against state-of-the-art approaches, significantly improving AUC and GAUC by over 0.02 and reducing logloss by over 13.7%. Further analyses demonstrate that DiseCTR successfully disentangles user interests, which is the key to OOD generalization for CTR prediction. We have released the code and data at https://github.com/DavyMorgan/DiseCTR/.