共词分析聚合研究领域的表面效度研究:基于自然语言处理

The Face Validity of Research Areas Identified by Co-words Analysis: Based on the Natural Language Process

为考察共词分析在研究领域聚合中的有效性,本研究以1988年～2017年三十年间发表且收录于WoS(Web of Science)数据库的计算机科学、地理学、数学、应用化学和应用物理学等五个学科的标题、摘要和关键词为分析对象,对不同分析单元下所聚合的研究领域的表面效度进行了检验。研究发现,关键词在研究领域聚合中具有相对最高的表面效度,摘要和题名的效度则相对较低。然而,本研究也发现,科学计量领域现有聚类指标和工具所聚合的研究领域缺乏区分能力且各种指标均存在着一定的效度风险。本研究的发现对于丰富科学计量领域关于研究领域聚合有效性的认识具有重要理论意义。

Aiming to test the face validity of research areas identified by co-words analysis, the present study selected the titles, ab- stracts and keywords from Computer Science, Geography, Mathematics, Applied Chemistry and Applied Physics which published with- in 1988 -2017 and collected into WOS （Web of Science） database as research subjects and tested the actual validity of research areas identified from different analysis units. The findings of this research include that the research area identified via keywords has highest face validity comparing to abstracts and titles. However, this study also found that there is a validity risk existing among varies indexes and tools in present scientific metrology research which aims to identify the research areas. The findings of this research have theoretical meanings for enriching the understanding of validity of research area identification in scientific metrology research.