본문 바로가기
  • Home

Analyzing Research Trends of Domestic Artificial Intelligence Research Using Network Analysis and Dynamic Topic Modelling

  • Journal of the Korean Society for Library and Information Science
  • 2021, 55(4), pp.141-157
  • DOI : 10.4275/KSLIS.2021.55.4.141
  • Publisher : 한국문헌정보학회
  • Research Area : Interdisciplinary Studies > Library and Information Science
  • Received : October 18, 2021
  • Accepted : November 15, 2021
  • Published : November 30, 2021

Woojin Jung 1 Chanhee Oh 1 Yong-Jun Zhu 2

1성균관대학교 문헌정보학과
2성균관대학교

Excellent Accredited

ABSTRACT

In this study, we aimed to understand research trends of domestic artificial intelligence research. To achieve the goal, we applied network analysis and dynamic topic modeling to domestic research papers on artificial intelligence. Among the papers that have been indexed in KCI (Korean Journal of Citation Index) by 2020, metadata and abstracts of 2,552 papers where the titles or indexed keywords include ‘artificial intelligence’ both in Korean and English were collected. Keyword, affiliation, subject field, and abstract were extracted and preprocessed for further analyses. We identified main keywords in the field by analyzing keyword co-occurrence networks as well as the degree and characteristics of research collaboration between domestic and foreign institutions and between industry and university by analyzing institutional collaboration networks. Dynamic topic modeling was performed on 1845 abstracts written in Korean, and 13 topics were obtained from the labeling process. This study broadens the understanding of domestic artificial intelligence research by identifying research trends through dynamic topic modeling from abstracts as well as the degree and characteristics of research collaboration through institutional collaboration networks from author affiliation information. In addition, the results of this study can be used by governmental institutions for making policies in accordance with artificial intelligence era.

Citation status

* References for papers published after 2023 are currently being built.