Topological Structure Learning Should Be A Research Priority for LLM-Based Multi-Agent Systems

TL;DR

This paper argues that the topology of LLM-based multi-agent systems should be a first-class design objective. It proposes a three-stage framework—agent selection, structure profiling, and topology synthesis—to learn dynamic, task-aware interaction structures that optimize performance under resource constraints. By formalizing MAS as directed graphs with macro archetypes and micro-level edge policies, it outlines concrete methods for agent selection, macro-structure mapping, and micro-topology synthesis, including counterfactual and TGNN-based approaches. The contribution highlights practical benefits in efficiency, robustness, and fairness, and identifies key challenges in benchmarking, scalability, and real-world deployment. Overall, the work provides a principled blueprint to advance adaptive, topology-aware MASs in real-world, complex tasks."

Abstract

Large Language Model-based Multi-Agent Systems (MASs) have emerged as a powerful paradigm for tackling complex tasks through collaborative intelligence. However, the topology of these systems--how agents in MASs should be configured, connected, and coordinated--remains largely unexplored. In this position paper, we call for a paradigm shift toward \emph{topology-aware MASs} that explicitly model and dynamically optimize the structure of inter-agent interactions. We identify three fundamental components--agents, communication links, and overall topology--that collectively determine the system's adaptability, efficiency, robustness, and fairness. To operationalize this vision, we introduce a systematic three-stage framework: 1) agent selection, 2) structure profiling, and 3) topology synthesis. This framework not only provides a principled foundation for designing MASs but also opens new research frontiers across language modeling, reinforcement learning, graph learning, and generative modeling to ultimately unleash their full potential in complex real-world applications. We conclude by outlining key challenges and opportunities in MASs evaluation. We hope our framework and perspectives offer critical new insights in the era of agentic AI.

Figures (2)

• Figure 1: Comparison of topological structures across datasets. Results show that various tasks may require a specific topological structure.
• Figure 2: The proposed three-stage framework: agent selection, structure profiling, and topology synthesis, which grounds four key positions.