Review articles as windows into Knowledge accumulation: the case of AI research

Abstract

Introduction. Review articles are essential in evolving scholarly information systems but have been underexplored in scientometrics. This paper aims to expand scientometric research on review articles, focusing on their role in understanding knowledge accumulation within specific domains.

Method. This study collected 4,315 review articles on artificial intelligence (AI). Using keyword frequency analysis and the Task-Technology Fit (TTF) model, the articles were classified into three categories: task-oriented, technology-oriented, and application-oriented.

Analysis. The temporal distribution of the review articles, the age distribution of their cited references, and the updating characteristics of references cited in review articles were analysed to provide preliminary insights into the evolution, dynamism, and updating patterns of knowledge in AI.

Results. The results show a marked increase in the publication frequency of review articles, especially over the past five years, with the application domain exhibiting the highest growth rate. Over half of the references cited in review articles across all domains are from the past five years. Additionally, older references are cited more frequently in newer reviews than more recent ones.

Conclusion. This study can be seen as an expansion of scientometric research based on review articles and highlights several intriguing research questions for exploration within this field.