Advances in Retrieval-Augmented Generation for Question Answering

The field of question answering is witnessing significant advancements with the integration of retrieval-augmented generation (RAG) techniques. Recent developments focus on enhancing the accuracy and faithfulness of generated answers by leveraging external knowledge sources, such as documents and knowledge graphs. The incorporation of multi-hop reasoning, dense retrieval, and advanced context fusion mechanisms has led to improved performance in complex question answering tasks. Additionally, researchers are exploring novel architectures, including multi-agent frameworks and dynamic vector stores, to optimize answer correctness and relevance. These innovations have far-reaching implications for real-world applications, particularly in enterprise settings and community question answering platforms. Noteworthy papers include: Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3, which presents a novel RAG framework for complex question answering tasks. RAGentA: Multi-Agent Retrieval-Augmented Generation for Attributed Question Answering, which introduces a multi-agent architecture for trustworthy answer generation. EraRAG: Efficient and Incremental Retrieval Augmented Generation for Growing Corpora, which proposes a novel multi-layered Graph-RAG framework for dynamic updates.

Sources

Enhancing Document-Level Question Answering via Multi-Hop Retrieval-Augmented Generation with LLaMA 3

eSapiens: A Real-World NLP Framework for Multimodal Document Understanding and Enterprise Knowledge Processing

Advancing Fact Attribution for Query Answering: Aggregate Queries and Novel Algorithms

RAGentA: Multi-Agent Retrieval-Augmented Generation for Attributed Question Answering

PersonalAI: Towards digital twins in the graph form

heiDS at ArchEHR-QA 2025: From Fixed-k to Query-dependent-k for Retrieval Augmented Generation

Inference Scaled GraphRAG: Improving Multi Hop Question Answering on Knowledge Graphs

Knowledge-Aware Diverse Reranking for Cross-Source Question Answering

EraRAG: Efficient and Incremental Retrieval Augmented Generation for Growing Corpora

ComRAG: Retrieval-Augmented Generation with Dynamic Vector Stores for Real-time Community Question Answering in Industry

Built with on top of