Integrating Clustering with Overlaps into Intelligent Agent Systems

Advances in Large Language Modeling (LLM) have allowed LLMs to be integrated into multi-agent problem-solving systems. We present a novel approach to enhance this technique and incorporate it, and some of its practical variants, into an intelligent agent system. Our approach is based on clustering with overlaps via an algorithm for Cluster Editing with Vertex Splitting (CEVS) to create a Network Hypothesis Search Agent. As a case study, we use a practical example of a panel discussion on possible mRNA treatments for Head and Neck cancer.