自然语言处理在我国社会科学领域应用的发展路径识别与构建研究

Research on the Identification of the Application Path of Natural Language Processingin Chinese Social Sciences

识别与构建自然语言处理在我国社会科学领域应用的发展路径有利于场景应用创新,对推动学科间深入融合与交叉发展具有重要意义。基于2010—2021年中国知网数据库中文社会科学引文索引(CSSCI)来源的期刊文献,利用CiteSpace软件并融合关键词阶段演进与中心度特征,开展自然语言处理技术在我国社科领域应用的发展路径识别研究,构建文本分类、机器翻译、情感分析、科学知识图谱4个研究方向发展路径。结果表明,文本分类、机器翻译、情感分析方向经历了由机器学习到深度学习、由独立模型到集成模型、由粗粒度任务到细粒度任务的演化进程,科学知识图谱方向的研究具有学科融合化、产业新兴性、政策理论化、方向创新性特点,但研究方法有待创新;自然语言处理与社会科学领域的持续融合与交叉,是其应用在我国社会科学领域持续发展的重要原因。

Identifying the application path of natural language processing in the field of Chinese social sciences is conducive to the innovation in scenario application, and is of significance to promote the in-depth integration and cross-development of disciplines. Based on the journal literature of CSSCI of CNKI database from 2010 to 2021, CiteSpace was used, and the characteristics of keywords stage evolution and centrality were combined, research on the identification of the application path of natural language processing in Chinese social sciences was carried out, de- velopment paths of four research directions were constructed, including text classification, machine translation, sentiment analysis and scientific knowledge map. It turns out that the directions of text classification, machine translation and sentiment analysis have experienced the evolution process from machine learning to deep learning, from independent model to integrated model, and from coarse-grained task to fine-grained task. The research in the direction of scientific knowledge map has the characteristics of discipline integration, industry emerging, policy theorization, and direction innovation.