Query-Aware Graph Neural Networks for Enhanced Retrieval-Augmented Generation
By: Vibhor Agrawal, Fay Wang, Rishi Puri
Potential Business Impact:
Helps computers answer tricky questions better.
We present a novel graph neural network (GNN) architecture for retrieval-augmented generation (RAG) that leverages query-aware attention mechanisms and learned scoring heads to improve retrieval accuracy on complex, multi-hop questions. Unlike traditional dense retrieval methods that treat documents as independent entities, our approach constructs per-episode knowledge graphs that capture both sequential and semantic relationships between text chunks. We introduce an Enhanced Graph Attention Network with query-guided pooling that dynamically focuses on relevant parts of the graph based on user queries. Experimental results demonstrate that our approach significantly outperforms standard dense retrievers on complex question answering tasks, particularly for questions requiring multi-document reasoning. Our implementation leverages PyTorch Geometric for efficient processing of graph-structured data, enabling scalable deployment in production retrieval systems
Similar Papers
Query-Specific GNN: A Comprehensive Graph Representation Learning Method for Retrieval Augmented Generation
Machine Learning (CS)
Helps AI answer harder questions by finding more facts.
Graph-Enhanced Retrieval-Augmented Question Answering for E-Commerce Customer Support
Computation and Language
Answers customer questions better using product facts.
Millions of $\text{GeAR}$-s: Extending GraphRAG to Millions of Documents
Computation and Language
Helps computers find better answers using connected facts.