Bibliometric Insights into Large Language Model-Driven Intelligent Agents: Themes, Trends and Prospects

Abstract

To clarify the research landscape and development trajectory of large language model-driven (LLM-driven) Intelligent agents, this study conducts a systematic bibliometric analysis of 620 literature. Key findings are as follows: (1) Globally, China, the U.S., and the U.K. are the most active in this field, contributing the largest publication volume and acting as hubs for technological innovation and academic collaboration. (2) Regarding publication carriers, IEEE Access, Applied Sciences-Basel, and Scientific Reports lead in publication volume; Journal of Medical Internet Research, Energy and Buildings, and IEEE Transactions on Visualization and Computer Graphics have the highest citation impact, reflecting their pivotal role in shaping the field. (3) Journal co-citation analysis identifies Advances In Neural Information Processing Systems, IEEE Access, IEEE Conference on Computer Vision and Pattern Recognition, Machine Learning, and Nature as central collaboration hubs, facilitating knowledge exchange and consolidating the core knowledge framework. (4) Keyword co-occurrence clustering reveals three research hotspots: fundamental architecture/technical foundations of LLM-driven agents, multi-agent collaboration mechanisms/system optimization, and domain-specific applications. (5) Keyword burst analysis points out three trends: advancement of agent embodied intelligence, enhancement of agent safety/trustworthiness/ethical governance, and interdisciplinary integration with emerging technologies. This study comprehensively maps the current state of LLM-driven agents research, providing valuable insights for the field of LLM-driven agents.

Ähnliche Arbeiten