A Long-term Value Prediction Framework In Video Ranking

2026-02-18

AI Summary

This work addresses three key challenges in long-term value (LTV) modeling for short-video recommendation ranking: position bias, attribution ambiguity, and temporal limitations. To tackle these issues, the authors propose a practical LTV prediction framework that incorporates a position-aware debiasing (PDQ) module to mitigate exposure position effects, replaces static attribution rules with a continuous multi-dimensional dynamic attribution mechanism, and employs a censoring-aware, day-granularity LTV objective to capture creator-driven long-term re-engagement. The framework is effectively integrated into existing ranking systems through quantile normalization, multi-dimensional signal fusion, a tailored hybrid loss function, and cross-temporal creator modeling. Both offline evaluations and online A/B experiments demonstrate significant improvements in LTV metrics, and the system has been stably deployed in Taobao’s large-scale production environment, consistently enhancing user engagement while maintaining alignment with short-term objectives.

Technology Category

• Machine Learning: Learning Preferences or Rankings
• Data Mining & Knowledge Management: Recommender Systems
• Search and Optimization: Algorithm Configuration

Application Category

• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Economics and fairness of platforms and recommendation systems

Abstract

Accurately modeling long-term value (LTV) at the ranking stage of short-video recommendation remains challenging. While delayed feedback and extended engagement have been explored, fine-grained attribution and robust position normalization at billion-scale are still underdeveloped. We propose a practical ranking-stage LTV framework addressing three challenges: position bias, attribution ambiguity, and temporal limitations. (1) Position bias: We introduce a Position-aware Debias Quantile (PDQ) module that normalizes engagement via quantile-based distributions, enabling position-robust LTV estimation without architectural changes. (2) Attribution ambiguity: We propose a multi-dimensional attribution module that learns continuous attribution strengths across contextual, behavioral, and content signals, replacing static rules to capture nuanced inter-video influence. A customized hybrid loss with explicit noise filtering improves causal clarity. (3) Temporal limitations: We present a cross-temporal author modeling module that builds censoring-aware, day-level LTV targets to capture creator-driven re-engagement over longer horizons; the design is extensible to other dimensions (e.g., topics, styles). Offline studies and online A/B tests show significant improvements in LTV metrics and stable trade-offs with short-term objectives. Implemented as task augmentation within an existing ranking model, the framework supports efficient training and serving, and has been deployed at billion-scale in Taobao's production system, delivering sustained engagement gains while remaining compatible with industrial constraints.

Problem

• long-term value
• video ranking
• position bias
• attribution ambiguity
• temporal limitations

Innovation

• Long-term Value Prediction
• Position-aware Debiasing
• Multi-dimensional Attribution
• Cross-temporal Modeling
• Video Ranking