The Determination and Visualisation of Key Concepts Related to the Training of Chatbots

This study aims to identify and visualize key concepts related to chatbot training through bibliometric analysis. The analysis of 549 sources from Scopus revealed a significant increase in publications from 2018, with a surge in 2023 likely driven by ChatGPT’s advent. We have identified four clusters of research areas. Those clusters are: (1) natural language processing; (2) application of natural language processing technologies in society; (3) application of machine learning for natural language processing; (4) chatbots in education and services. Central concepts were identified within each cluster. The results of our findings define natural language understanding, language modelling, controlled use of large language models in education, application of virtual assistants and diagnostic systems, and integration of chatbots into adaptive learning systems as the most prominent leading research directions. The same results offer implications for education, AI research, and organizational strategies for integrating conversational agents. Key concepts are possible to integrate into curriculum development and future research in natural language processing.