Comment Staytime Prediction with LLM-enhanced Comment Understanding

2025-04-02

AI Summary

This work addresses the novel task of predicting user dwell time in the comment sections of short-video platforms—a problem overlooked by existing video watch-time models, which ignore comment-level fine-grained interactions and their semantic associations. To bridge this gap, we introduce KuaiComt, the first micro-video recommendation dataset with fine-grained comment annotations. We further propose the LLM-enhanced Comment Understanding (LCU) framework: (1) formally defining comment-level dwell time prediction; (2) designing a two-stage LLM adaptation mechanism for cross-modal semantic alignment between videos and comments; and (3) incorporating a dual-comment ranking auxiliary task to model user preference toward individual comments. Our method integrates domain-adaptive LLM fine-tuning, multi-task learning, and comment sequence modeling. Offline experiments demonstrate significant improvements in prediction accuracy. Online A/B testing confirms substantial gains in both user comment engagement rate and overall dwell time.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Computer Vision: Video Understanding & Activity Analysis
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

In modern online streaming platforms, the comments section plays a critical role in enhancing the overall user experience. Understanding user behavior within the comments section is essential for comprehensive user interest modeling. A key factor of user engagement is staytime, which refers to the amount of time that users browse and post comments. Existing watchtime prediction methods struggle to adapt to staytime prediction, overlooking interactions with individual comments and their interrelation. In this paper, we present a micro-video recommendation dataset with video comments (named as KuaiComt) which is collected from Kuaishou platform. correspondingly, we propose a practical framework for comment staytime prediction with LLM-enhanced Comment Understanding (LCU). Our framework leverages the strong text comprehension capabilities of large language models (LLMs) to understand textual information of comments, while also incorporating fine-grained comment ranking signals as auxiliary tasks. The framework is two-staged: first, the LLM is fine-tuned using domain-specific tasks to bridge the video and the comments; second, we incorporate the LLM outputs into the prediction model and design two comment ranking auxiliary tasks to better understand user preference. Extensive offline experiments demonstrate the effectiveness of our framework, showing significant improvements on the task of comment staytime prediction. Additionally, online A/B testing further validates the practical benefits on industrial scenario. Our dataset KuaiComt (https://github.com/lyingCS/KuaiComt.github.io) and code for LCU (https://github.com/lyingCS/LCU) are fully released.

Problem

• Predict user staytime in video comment sections
• Enhance comment understanding using LLMs
• Improve user engagement modeling with fine-grained signals

Innovation

• LLM-enhanced comment understanding for staytime prediction
• Two-staged framework with fine-tuned LLM
• Auxiliary tasks for fine-grained comment ranking