Detecting the knowledge structure of generative artificial intelligence research

Since the introduction of ChatGPT, generative artificial intelligence (GenAI) has attracted significant attention in the academic community, leading to a surge in the publication of GenAI papers. It is essential to analyse and explore the research landscape of this growing body of literature. Unlike conventional literature analysis methods such as bibliometric analysis, this study utilises text mining and knowledge networks to offer a comprehensive review of the current knowledge structure in GenAI. To achieve this, we applied latent Dirichlet allocation to identify 15 research topics from 18,459 GenAI papers. In addition, we created knowledge networks for each research topic based on the thematic proximity of pairwise papers, enabling us to assess research concentration in each topic using four network metrics. Our research not only enhances understanding of the evolution of GenAI research but also expands the application of text mining and knowledge networks in literature analysis.