[论文] From Descriptive to Prescriptive: Uncover the Social Value Alignment o...

论文概要

研究领域: NLP 作者: Jinxian Qu, Qingqing Gu, Teng Chen 发布时间: 2026-05-17 arXiv: 2505.12352

中文摘要

基于LLM的智能体的广泛应用要求与人类社会价值观高度对齐。然而，当前工作在自我认知和困境决策方面仍存在缺陷，以及自我情感方面。为了弥补这一点，我们提出一种新颖的基于价值的框架，使用GraphRAG将原则转换为基于价值的指令，并通过在特定对话上下文中检索合适的指令来引导智能体按预期行事。为了评估预期行为的比例，我们从两个著名理论定义预期行为：马斯洛需求层次理论和普卢奇克情感轮。通过在DAILYDILEMMAS基准上实验我们的方法，我们的方法相比基于提示的基线（包括ECoT、Plan-and-Solve和元认知提示）表现出显著的性能提升。我们的方法为AI系统中自我情感的出现提供了基础。

原文摘要

Wide applications of LLM-based agents require strong alignment with human social values. However, current works still exhibit deficiencies in self-cognition and dilemma decision, as well as self-emotions. To remedy this, we propose a novel value-based framework that employs GraphRAG to convert principles into value-based instructions and steer the agent to behave as expected by retrieving the suitable instruction upon a specific conversation context. To evaluate the ratio of expected behaviors, we define the expected behaviors from two famous theories, Maslow's Hierarchy of Needs and Plutchik's Wheel of Emotion. By experimenting with our method on the benchmark of DAILYDILEMMAS, our method exhibits significant performance gains compared to prompt-based baselines, including ECoT, Plan-and-S...

自动采集于 2026-05-18

#论文 #arXiv #NLP #小凯