Large Language Models Empowered Personalized Web Agents

Hongru Cai , Yongqi Li , Wenjie Wang , Fengbin Zhu , Xiaoyu Shen , Wenjie Li , Tat-Seng Chua

🏛 Institutions: NUS , PolyU , USTC , Eastern Institute of Technology , Ningbo
📅 Date: October 22, 2024
📑 Publisher: WWW 2025
💻 Env: Web
🔑 Keywords: benchmark personalization PUMA PersonalWAB memory bank

TLDR

This paper formulates personalized web agents that condition on user profiles and historical web behaviors, introduces the PersonalWAB benchmark for evaluating that setting, and proposes the PUMA alignment method. PUMA uses a memory bank with task-specific retrieval plus fine-tuning and preference optimization to improve user-dependent action execution.

Open paper Report issue