User Behavior Simulation with Large Language Model based Agents
Lei Wang Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Jingsen Zhang Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Hao Yang Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Zhi-Yuan Chen Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Jiakai Tang Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Zeyu Zhang Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Xu Chen Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Yankai Lin Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Hao Sun Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Ruihua Song Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Wayne Xin Zhao Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Jun Xu Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Zhicheng Dou Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China Jun Wang University College London, London, UK Ji-Rong Wen Gaoling School of Artificial Intelligence, Renmin University of China, Beijing, China Beijing Key Laboratory of Big Data Management and Analysis Methods, Beijing, China
Abstract
Simulating high quality user behavior data has always been a fundamental problem in human-centered applications, where the major difficulty originates from the intricate mechanism of human decision process. Recently, substantial evidences have suggested that by learning huge amounts of web knowledge, large language models (LLMs) can achieve human-like intelligence. We believe these models can provide significant opportunities to more believable user behavior simulation. To inspire such direction, we propose an LLM-based agent framework and design a sandbox environment to simulate real user behaviors. Based on extensive experiments, we find that the simulated behaviors of our method are very close to the ones of real humans. Concerning potential applications, we simulate and study two social phenomenons including (1) information cocoons and (2) user conformity behaviors. This research provides novel simulation paradigms for human-centered applications.
中文速览
真实用户行为数据难以大量获取，而现有的用户行为模拟方法又过度依赖真实数据、决策模型过于简化、且只能局限于单一场景。为此，研究者提出了一套基于大语言模型（LLM）的智能体模拟框架 RecAgent，为每位用户构建一个具备个人档案、记忆（感觉记忆、短期记忆、长期记忆）和行动模块的智能体，并搭建了一个可干预、可重置的沙盒环境，让智能体在推荐系统中浏览、搜索、点击内容，同时与其他智能体进行一对一聊天或一对多广播。大量实验表明，RecAgent 模拟的用户行为比传统方法平均提升约 68%，与真实人类行为的差距仅约 8%，在聊天和广播场景下无需针对特定数据集微调也能产生可信行为。研究者进一步用该框架复现并深入分析了"信息茧房"和"用户从众"两种社会现象，并探索了缓解这些现象的潜在策略——这表明以 LLM 为核心的用户行为模拟，有望成为以人为中心的 AI 应用研究的全新范式。
原文 arXiv:2306.02552;中英对照 + 大白话阅读 https://aha.fim.ai/paper/2306.02552v3