Hierarchical Formation of Causal Networks Based on ChatGPT

Abstract

This paper is devoted to a methodology of forming causal networks by applying the ChatGPT system repeatedly, and visualizing and analyzing these networks with Gephi. The methodology is based on the use of the ChatGPT system, a generative pre-trained transformer on large text corpora, which uses artificial intelligence capabilities to perform user prompts. The methodology covers the means of analysis and visualization of the formed networks using the Gephi program. The CSV format is used to upload data to the Gephi environment. The article shows the possibility of constructing causal networks of concepts based on the use of Chat GPT, which allows for solving problems that previously required the involvement of large resources (human and time). The methodology integrates means of intellectual text analytics and network analysis, as well as their visualization. The formed causal networks provide the possibility of further transition to scenario analysis. The article discusses the possibility of emulating a multitude of experts by repeatedly applying similar prompts to the ChatGPT system. The proposed comprehensive methodology can be applied to the construction of causal networks in various subject areas.

Keywords: Chat GPT, causal networks, Domain model, Artificial experts, Graph visualization, Cyber Security