NG-Router: Graph-Supervised Multi-Agent Collaboration for Nutrition Question Answering

2025-10-10

AI Summary

Nutritional question-answering systems face two key challenges: limited reasoning capacity of single-agent architectures and poor decision accuracy in multi-agent systems due to architectural complexity and context overload. To address these, we propose NG-Router—a knowledge graph-guided multi-agent collaborative framework. It embeds agent nodes into a heterogeneous knowledge graph, employs graph neural networks to learn task-aware agent routing distributions, and introduces a gradient-driven subgraph retrieval mechanism to enhance multi-hop and relational reasoning. Furthermore, it leverages empirical performance to generate soft supervision signals for optimizing collaborative decision-making. Evaluated across multiple nutritional health benchmark datasets and diverse backbone models, NG-Router consistently outperforms both single-agent baselines and ensemble methods. Results demonstrate its effectiveness and robustness in complex, personalized dietary guidance tasks.

Technology Category

• Multiagent Systems: Adversarial Agents
• Cognitive Modeling & Cognitive Systems: Agent Architectures
• Natural Language Processing: Safety and Robustness

Application Category

• Search and Retrieval-Augmented AI: Agentic search
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Diet plays a central role in human health, and Nutrition Question Answering (QA) offers a promising path toward personalized dietary guidance and the prevention of diet-related chronic diseases. However, existing methods face two fundamental challenges: the limited reasoning capacity of single-agent systems and the complexity of designing effective multi-agent architectures, as well as contextual overload that hinders accurate decision-making. We introduce Nutritional-Graph Router (NG-Router), a novel framework that formulates nutritional QA as a supervised, knowledge-graph-guided multi-agent collaboration problem. NG-Router integrates agent nodes into heterogeneous knowledge graphs and employs a graph neural network to learn task-aware routing distributions over agents, leveraging soft supervision derived from empirical agent performance. To further address contextual overload, we propose a gradient-based subgraph retrieval mechanism that identifies salient evidence during training, thereby enhancing multi-hop and relational reasoning. Extensive experiments across multiple benchmarks and backbone models demonstrate that NG-Router consistently outperforms both single-agent and ensemble baselines, offering a principled approach to domain-aware multi-agent reasoning for complex nutritional health tasks.

Problem

• Addresses limited reasoning in single-agent nutrition QA systems
• Solves complex multi-agent architecture design for dietary guidance
• Mitigates contextual overload hindering accurate nutritional decision-making

Innovation

• Graph-supervised multi-agent collaboration for nutrition QA
• GNN-based routing distribution over agent nodes
• Gradient-based subgraph retrieval for salient evidence