数字人文视域下研究方法的演变:基于CNKI数据与大语言模型的量化分析  

The Evolution of Research Methods in the Digital Humanities Perspective:A Quantitative Analysis Based on CNKI Data and a Large Language Model

作　　者：孙光耀 王东波[1,2] SUN Guangyao;WANG Dongbo(School of Information Management,Nanjing Agricultural University,Nanjing 210095;Research Center for Humanities and Social Computing,Nanjing Agricultural University,Nanjing 210095)

机构地区：[1]南京农业大学信息管理学院,南京210095 [2]南京农业大学人文与社会计算研究中心,南京210095

出　　处：《科技情报研究》2025年第1期65-74,共10页Scientific Information Research

基　　金：2021年国家社会科学基金重大项目“中国古代典籍跨语言知识库构建及应用研究”(编号:21&ZD331)。

摘　　要：[目的/意义]文章旨在利用大语言模型技术,探索数字人文领域研究方法的演变趋势。[方法/过程]文章主要以CNKI期刊论文数据为研究对象,选取通用中文大语言模型GLM-4,采用提示词工程、思维链,对论文摘要数据进行抽取、聚类,并通过量化处理分析其演变趋势。[结果/结论]研究表明,GLM-4能够很好地从复杂的摘要数据中识别并抽取出研究方法。按时序变化分析演变趋势,发现“访谈调研”“扎根理论”等研究方法逐渐边缘化,机器学习等相关研究方法逐步成为主流。文章揭示了中文数字人文领域研究方法的演变趋势,赋予数字人文的研究成果更为丰富且全面的文化内涵。[Purpose/significance]This paper aims to explore the evolution trend of research methods in the field of digital humanities with the help of large language model technology . [Method/process]This paper mainly focuses on the data of CNKI journal articles, selects the general Chinese large language model GLM-4, uses prompt engineering and chain of thought to extract and cluster the abstract data, of papers and analyzes its evolution trend through quantitative processing. [Result/conclusion]The study shows that GLM-4 can well identify and extract research methods from complex abstract data. Analyzing the evolution trend in chronological order, it is found that research methods such as "interview survey" and "grounded theory" are gradually marginalized, while machine learning and other related research methods are gradually becoming mainstream. This article reveals the evolution trend of research methods in the field of Chinese digital humanities, and gives the research results of digital humanities a richer and more comprehensive cultural connotation.

关 键 词：数字人文 大语言模型 GLM-4 提示词工程 思维链 机器学习 

分 类 号：G353.1[文化科学—情报学] TP391.1[自动化与计算机技术—计算机应用技术]