Leveraging Watch-time Feedback for Short-Video Recommendations: A Causal Labeling Framework

TL;DR

This paper introduces a causal, multi-semantics labeling framework for short-video recommendation systems that effectively leverages watch time feedback, mitigating bias and improving user engagement.

Contribution

The paper proposes DML, a novel framework that constructs multi-semantics labels from watch time using quantiles and causal adjustment to reduce bias in short-video recommendations.

Findings

DML improves recommendation accuracy in offline tests.

DML increases user engagement in online deployment.

Effective bias mitigation at label level enhances user interest modeling.

Abstract

With the proliferation of short video applications, the significance of short video recommendations has vastly increased. Unlike other recommendation scenarios, short video recommendation systems heavily rely on feedback from watch time. Existing approaches simply treat watch time as a direct label, failing to effectively harness its extensive semantics and introduce bias, thereby limiting the potential for modeling user interests based on watch time. To overcome this challenge, we propose a framework named Debiased Multiple-semantics-extracting Labeling(DML). DML constructs labels that encompass various semantics by utilizing quantiles derived from the distribution of watch time, prioritizing relative order rather than absolute label values. This approach facilitates easier model learning while aligning with the ranking objective of recommendations. Furthermore, we introduce a method inspired by causal adjustment to refine label definitions, thereby directly mitigating bias at the label level. We substantiate the effectiveness of our DML framework through both online and offline experiments. Extensive results demonstrate that our DML could effectively leverage watch time to discover users' real interests, enhancing their engagement in our application.