Pre-Trained Language Models as Zero-Shot Tools for Social Science Research

查看原文 / View Original 编辑 / Edit

Computer Science > Computation and Language arXiv:2610.19048 (cs) [Submitted on 13 Oct 2026] Title:Pre-Trained Language Models as Zero-Shot Tools for Social Science Research Authors:Noah Dasanaike View a PDF of the paper titled Pre-Trained Language Models as Zero-Shot Tools for Social Science Research, by Noah Dasanaike View PDF HTML (experimental)

计算机科学 > 计算与语言 arXiv:2610.19048 (cs) [提交于 2026 年 10 月 13 日] 标题：预训练语言模型作为社会科学研究的零样本工具作者：Noah Dasanaike 查看标题为《预训练语言模型作为社会科学研究的零样本工具》的论文 PDF，作者：Noah Dasanaike 查看 PDF HTML（实验性）

Abstract:Record linkage and demographic inference from names are routine demands in social science. Most approaches to these problems require labeled training data that are difficult or costly to acquire. I show that pre-trained language models perform both tasks accurately without requiring task-specific training.

摘要：记录链接和基于姓名的人口统计推断是社会科学中的常规需求。解决这些问题的大多数方法都需要难以获取或成本高昂的标记训练数据。我证明了预训练语言模型无需特定任务的训练即可准确执行这两项任务。

For record linkage, I introduce EnsembleLink, an algorithm that uses pre-trained embeddings alongside string-based scorers to retrieve a short list of candidates and then identify the likeliest matches. EnsembleLink matches or exceeds the performance of most prior approaches while generalizing across scripts and formats that those methods are unable to handle.

在记录链接方面，我引入了 EnsembleLink，这是一种结合使用预训练嵌入和基于字符串的评分器来检索候选列表，进而识别最可能匹配项的算法。EnsembleLink 的性能与大多数先前的方法相当或更优，同时能够推广到那些方法无法处理的各种脚本和格式中。

For ethnicity inference from names, I show that large language models (LLMs) achieve up to 83.8% accuracy on U.S. voter files, compared to 78.0% for name-supplemented BISG. Open-weight models offer reproducible implementations that approach these models in accuracy.

在基于姓名的种族推断方面，我证明了大型语言模型（LLMs）在美国选民文件上的准确率高达 83.8%，而补充了姓名的 BISG 准确率为 78.0%。开放权重模型提供了可复现的实现，其准确率接近这些模型。

Moreover, LLMs are able to infer group membership for religious, caste-based, and other ethnic group categories outside of the United States, and reduce bias in downstream estimates of geographic racial gaps. Where pre-training encodes the world knowledge required by each task, task-specific labels add less information.

此外，LLMs 能够推断美国以外地区宗教、种姓及其他族群类别的成员身份，并减少在地理种族差距下游估计中的偏差。在预训练已经编码了每项任务所需的世界知识的情况下，特定任务的标签所增加的信息量较少。