From Bootstrapping to Sequence Modeling: A Unified Generative Framework for Personalized Landing-Page Modeling

2026-06-26

AI Summary

Existing CQL-based personalized landing page recommendation methods are constrained by the Markov assumption, struggling to capture non-Markovian temporal dependencies in user behavior and facing challenges in credit assignment and error accumulation under delayed rewards and long-horizon scenarios. This work proposes GLAN, a novel framework that introduces sequence modeling into this task for the first time by building upon the Decision Transformer to establish a unified global-local paradigm. Specifically, the L-RTG module captures cross-day consumption dynamics to provide long-term return guidance, while the HRM module decomposes sparse session-level feedback into fine-grained supervisory signals. By circumventing conventional reinforcement learning limitations, GLAN enables end-to-end sequential decision generation. Experiments on the Kuaishou platform demonstrate significant improvements in key metrics, with a 0.158% increase in daily active users (DAU) and a 0.108% extension in user lifetime (LT).

Abstract

Modern online platforms increasingly adopt multi-page architectures to accommodate diverse user needs. On these platforms, page navigation (the process of directing users to specific functional pages upon app entry) serves as a critical gateway that shapes user's first impression and significantly influences subsequent engagement. To optimize this process, Kuaishou formulated the task of Personalized Landing Page Modeling (PLPM) and proposed KLAN, a reinforcement learning framework built upon Conservative Q-Learning (CQL). However, CQL-based approaches suffer from two fundamental limitations: (1) the Markov assumption fails to capture the strong non-Markovian temporal dependencies inherent in real-world user behaviors, and (2) TD learning with bootstrapping incurs severe cumulative errors and credit assignment difficulties under delayed rewards, particularly in long-horizon settings where users enter the app multiple times daily. To address these limitations, we propose GLAN (Generative Landing-page Adaptive Navigator), a sequence modeling framework built on Decision Transformer to tackle PLPM from a unified global-local perspective. Specifically, GLAN incorporates two key modules. First, we design the L-RTG module that captures users' inter-day consumption dynamics to provide accurate global guidance for all page assignments within a day. Furthermore, we propose the HRM module that decomposes session-level feedback into fine-grained signals, enabling precise local supervision for each page assignment. Extensive online experiments conducted on the Kuaishou platform demonstrate the effectiveness of GLAN, achieving +0.158\% and +0.108\% improvements on Daily Active Users (DAU) and user Lifetime (LT) respectively.

Problem

• Personalized Landing Page Modeling
• Non-Markovian Dependencies
• Delayed Rewards
• Credit Assignment
• Bootstrapping Error

Innovation

• Sequence Modeling
• Decision Transformer
• Personalized Landing Page
• Non-Markovian Dynamics
• Global-Local Supervision