ORCA: Mitigating Over-Reliance for Multi-Task Dwell Time Prediction with Causal Decoupling

2025/08/22 by Luo, Huishi, Zhuang, Fuzhen, Zhu, Yongchun +6
#FOS: Computer and information sciences #Information Retrieval (cs.IR)

paper · doi:10.48550/arxiv.2508.16573

Abstract

Dwell time (DT) is a critical post-click metric for evaluating user preference in recommender systems, complementing the traditional click-through rate (CTR). Although multi-task learning is widely adopted to jointly optimize DT and CTR, we observe that multi-task models systematically collapse their DT predictions to the shortest and longest bins, under-predicting the moderate durations. We attribute this moderate-duration bin under-representation to over-reliance on the CTR-DT spurious correlation, and propose ORCA to address it with causal-decoupling. Specifically, ORCA explicitly models and subtracts CTR's negative transfer while preserving its positive transfer. We further introduce (i) feature-level counterfactual intervention, and (ii) a task-interaction module with instance inverse-weighting, weakening CTR-mediated effect and restoring direct DT semantics. ORCA is model-agnostic and easy to deploy. Experiments show an average 10.6% lift in DT metrics without harming CTR. Code is available at https://github.com/Chrissie-Law/ORCA-Mitigating-Over-Reliance-for-Multi-Task-Dwell-Time-Prediction-with-Causal-Decoupling.

Citations

• One for Dozens: Adaptive REcommendation for All Domains with Counterfactual Augmentation
• Counteracting Duration Bias in Video Recommendation via Counterfactual Watch Time
• Tenrec: A Large-scale Multipurpose Benchmark Dataset for Recommender Systems
• Deconfounding Duration Bias in Watch-time Prediction for Video Recommendation
• Achieving Counterfactual Fairness for Causal Bandit
• Dependent Multi-Task Learning with Causal Intervention for Image Captioning
• Modeling the Sequential Dependence among Audience Multi-step Conversions with Multi-task Learning in Targeted Display Advertising
• Adaptive Factorization Network: Learning Adaptive-Order Feature Interactions
• Entire Space Multi-Task Model: An Effective Approach for Estimating Post-Click Conversion Rate
• Neural Collaborative Filtering
• DeepFM: A Factorization-Machine based Neural Network for CTR Prediction

Related

• Time Matters: Enhancing Pre-trained News Recommendation Models with Robust User Dwell Time Injection
• Distillation based Multi-task Learning: A Candidate Generation Model for Improving Reading Duration
• CTR-Sink: Attention Sink for Language Models in Click-Through Rate Prediction
• Entire Space Counterfactual Learning: Tuning, Analytical Properties and Industrial Applications
• DCMT: A Direct Entire-Space Causal Multi-Task Framework for Post-Click Conversion Estimation