MME-RAG: Multi-Manager-Expert Retrieval-Augmented Generation for Fine-Grained Entity Recognition in Task-Oriented Dialogues

TL;DR

MME-RAG introduces a hierarchical, retrieval-augmented framework for fine-grained entity recognition in task-oriented dialogues, improving domain adaptation and robustness without extra training.

Contribution

It proposes a novel multi-manager-expert retrieval-augmented generation approach with hierarchical decomposition and retrieval support for better domain adaptation.

Findings

Outperforms recent baselines in multiple domains.

Hierarchical decomposition improves robustness.

KeyInfo retrieval enhances cross-domain generalization.

Abstract

Fine-grained entity recognition is crucial for reasoning and decision-making in task-oriented dialogues, yet current large language models (LLMs) continue to face challenges in domain adaptation and retrieval controllability. We introduce MME-RAG, a Multi-Manager-Expert Retrieval-Augmented Generation framework that decomposes entity recognition into two coordinated stages: type-level judgment by lightweight managers and span-level extraction by specialized experts. Each expert is supported by a KeyInfo retriever that injects semantically aligned, few-shot exemplars during inference, enabling precise and domain-adaptive extraction without additional training. Experiments on CrossNER, MIT-Movie, MIT-Restaurant, and our newly constructed multi-domain customer-service dataset demonstrate that MME-RAG performs better than recent baselines in most domains. Ablation studies further show that both the hierarchical decomposition and KeyInfo-guided retrieval are key drivers of robustness and cross-domain generalization, establishing MME-RAG as a scalable and interpretable solution for adaptive dialogue understanding.