Shop-R1: Rewarding LLMs to Simulate Human Behavior in Online Shopping via Reinforcement Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Zhang, Yimeng, Wang, Tian, Gesi, Jiri, Wang, Ziyi, Lu, Yuxuan, Lin, Jiacheng, Zhan, Sinong, Gao, Vianne, Jiao, Ruochen, Liu, Junze, Qian, Kun, Tang, Yuxin, Xue, Ran, Zhang, Houyu, Cui, Qingjun, Guo, Yufan, Wang, Dakuo
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910029546782720
author Zhang, Yimeng
Wang, Tian
Gesi, Jiri
Wang, Ziyi
Lu, Yuxuan
Lin, Jiacheng
Zhan, Sinong
Gao, Vianne
Jiao, Ruochen
Liu, Junze
Qian, Kun
Tang, Yuxin
Xue, Ran
Zhang, Houyu
Cui, Qingjun
Guo, Yufan
Wang, Dakuo
author_facet Zhang, Yimeng
Wang, Tian
Gesi, Jiri
Wang, Ziyi
Lu, Yuxuan
Lin, Jiacheng
Zhan, Sinong
Gao, Vianne
Jiao, Ruochen
Liu, Junze
Qian, Kun
Tang, Yuxin
Xue, Ran
Zhang, Houyu
Cui, Qingjun
Guo, Yufan
Wang, Dakuo
contents Large Language Models (LLMs) have recently demonstrated strong potential in generating 'believable human-like' behavior in web environments. Prior work has explored augmenting training data with LLM-synthesized rationales and applying supervised fine-tuning (SFT) to enhance reasoning ability, which in turn can improve downstream action prediction. However, the performance of such approaches remains inherently bounded by the reasoning capabilities of the model used to generate the rationales. In this paper, we introduce Shop-R1, a novel reinforcement learning (RL) framework aimed at enhancing the reasoning ability of LLMs for simulation of real human behavior in online shopping environments. Specifically, Shop-R1 decomposes the human behavior simulation task into two stages: rationale generation and action prediction, each guided by distinct reward signals. For rationale generation, we leverage internal model signals (e.g., logit distributions) to guide the reasoning process in a self-supervised manner. For action prediction, we propose a hierarchical reward structure with difficulty-aware scaling to prevent reward hacking and enable fine-grained reward assignment. This design evaluates both high-level action types and the correctness of fine-grained sub-action details (attributes and values), rewarding outputs proportionally to their difficulty. Experimental results show that our method achieves a relative improvement of over 65% compared to the baseline. The project page is available at https://damon-demon.github.io/shop-r1.html.
format Preprint
id arxiv_https___arxiv_org_abs_2507_17842
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Shop-R1: Rewarding LLMs to Simulate Human Behavior in Online Shopping via Reinforcement Learning
Zhang, Yimeng
Wang, Tian
Gesi, Jiri
Wang, Ziyi
Lu, Yuxuan
Lin, Jiacheng
Zhan, Sinong
Gao, Vianne
Jiao, Ruochen
Liu, Junze
Qian, Kun
Tang, Yuxin
Xue, Ran
Zhang, Houyu
Cui, Qingjun
Guo, Yufan
Wang, Dakuo
Computation and Language
Large Language Models (LLMs) have recently demonstrated strong potential in generating 'believable human-like' behavior in web environments. Prior work has explored augmenting training data with LLM-synthesized rationales and applying supervised fine-tuning (SFT) to enhance reasoning ability, which in turn can improve downstream action prediction. However, the performance of such approaches remains inherently bounded by the reasoning capabilities of the model used to generate the rationales. In this paper, we introduce Shop-R1, a novel reinforcement learning (RL) framework aimed at enhancing the reasoning ability of LLMs for simulation of real human behavior in online shopping environments. Specifically, Shop-R1 decomposes the human behavior simulation task into two stages: rationale generation and action prediction, each guided by distinct reward signals. For rationale generation, we leverage internal model signals (e.g., logit distributions) to guide the reasoning process in a self-supervised manner. For action prediction, we propose a hierarchical reward structure with difficulty-aware scaling to prevent reward hacking and enable fine-grained reward assignment. This design evaluates both high-level action types and the correctness of fine-grained sub-action details (attributes and values), rewarding outputs proportionally to their difficulty. Experimental results show that our method achieves a relative improvement of over 65% compared to the baseline. The project page is available at https://damon-demon.github.io/shop-r1.html.
title Shop-R1: Rewarding LLMs to Simulate Human Behavior in Online Shopping via Reinforcement Learning
topic Computation and Language
url https://arxiv.org/abs/2507.17842