A bibliometric analysis of large language models in China

Purpose This paper aims to review the development path of Large Language Models (LLMs) and summarize the research hotspots and frontiers of LLMs in China so as to provide suggestions for further optimization in this field. Design/methodology/approach This study employs bibliometric analysis using CiteSpace software. The data were extracted from CNKI using an advanced search with keywords, limiting sources to journals, and refining the dataset to relevant articles. A total of 751 Chinese research papers on LLMs from 2021 to 2024 were retrieved and selected. Co-occurrence and clustering diagrams were used to analyze the publication trends in journals, keywords, authors, and research institutions through integrating the Antecedents-Decisions-Outcomes (ADO) and Theory-Context-Characteristic-Method (TCCM) frameworks. Findings It is found that the research on LLMs in China is still in an early and rapidly developing stage, with applications in fields such as education, medicine, and finance. The results indicate that the number of academic publications on LLMs was set to surge after 2023 due to the advancements of ChatGPT. The research focused on topics like artificial intelligence, human-computer collaboration, and ethical frameworks. Originality/value This paper provides an in-depth exploration at the academic prospects of LLMs in China, gaining insights into its evolving research focuses and potential applications. The importance of balancing technological progress and ethical considerations in practical applications is also emphasized.