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

TL;DR

This paper introduces ORCA, a causal decoupling method that improves multi-task dwell time prediction by reducing over-reliance on CTR-DT correlation, leading to more accurate moderate-duration predictions.

Contribution

ORCA is a novel causal decoupling approach that explicitly models and subtracts CTR's negative transfer, enhancing dwell time prediction in recommender systems.

Findings

10.6% average improvement in dwell time metrics

Preserves CTR performance while improving DT predictions

Model-agnostic and easy to deploy

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.