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

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Luo, Huishi, Zhuang, Fuzhen, Zhu, Yongchun, Wu, Yiqing, Kang, Bo, Xie, Ruobing, Xia, Feng, Wang, Deqing, Dong, Jin
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918129143119872
author Luo, Huishi
Zhuang, Fuzhen
Zhu, Yongchun
Wu, Yiqing
Kang, Bo
Xie, Ruobing
Xia, Feng
Wang, Deqing
Dong, Jin
author_facet Luo, Huishi
Zhuang, Fuzhen
Zhu, Yongchun
Wu, Yiqing
Kang, Bo
Xie, Ruobing
Xia, Feng
Wang, Deqing
Dong, Jin
contents 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.
format Preprint
id arxiv_https___arxiv_org_abs_2508_16573
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle ORCA: Mitigating Over-Reliance for Multi-Task Dwell Time Prediction with Causal Decoupling
Luo, Huishi
Zhuang, Fuzhen
Zhu, Yongchun
Wu, Yiqing
Kang, Bo
Xie, Ruobing
Xia, Feng
Wang, Deqing
Dong, Jin
Information Retrieval
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.
title ORCA: Mitigating Over-Reliance for Multi-Task Dwell Time Prediction with Causal Decoupling
topic Information Retrieval
url https://arxiv.org/abs/2508.16573