用户模拟中需要基于社会的个性框架
The Need for a Socially-Grounded Persona Framework for User Simulation
AI总结：
本文提出SCOPE框架，通过社会心理学协议构建更高质量的个性，提升大语言模型在社会模拟中的表现。
摘要：
合成个性广泛用于调节大语言模型（LLMs）进行社会模拟，但大多数个性仍由粗略的社会人口统计数据或摘要构建。我们重新审视个性创建，引入SCOPE，一个基于社会的个性构建和评估框架，该框架基于从124名美国参与者收集的141项、两小时的社会心理学协议。在七个模型中，我们发现仅基于人口统计的个性是结构瓶颈：人口统计仅能解释人类响应相似性变异的约1.5%。添加社会心理学特征可以提高行为预测并减少过度强调，而非人口统计的基于价值观和身份的个性在显著降低偏见的情况下实现了强对齐。这些趋势在SimBench（441个对齐问题）中得到推广，SCOPE个性在默认提示和NVIDIA Nemotron个性上表现更优，而SCOPE增强提高了基于Nemotron的个性。我们的结果表明，个性质量取决于社会心理学结构，而非人口统计模板或摘要。
英文摘要：
Synthetic personas are widely used to condition large language models (LLMs) for social simulation, yet most personas are still constructed from coarse sociodemographic attributes or summaries. We revisit persona creation by introducing SCOPE, a socially grounded framework for persona construction and evaluation, built from a 141-item, two-hour sociopsychological protocol collected from 124 U.S.-based participants. Across seven models, we find that demographic-only personas are a structural bottleneck: demographics explain only ~1.5% of variance in human response similarity. Adding sociopsychological facets improves behavioral prediction and reduces over-accentuation, and non-demographic personas based on values and identity achieve strong alignment with substantially lower bias. These trends generalize to SimBench (441 aligned questions), where SCOPE personas outperform default prompting and NVIDIA Nemotron personas, and SCOPE augmentation improves Nemotron-based personas. Our results indicate that persona quality depends on sociopsychological structure rather than demographic templates or summaries.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
