The field of retrieval-augmented generation (RAG) is rapidly advancing, with a focus on improving the accuracy and efficiency of knowledge retrieval and generation. Recent developments have seen the introduction of novel graph neural network architectures, such as query-aware graph neural networks, which leverage query-aware attention mechanisms and learned scoring heads to improve retrieval accuracy. Additionally, there is a growing interest in applying RAG to specific domains, such as the electric power industry, where it can be used to build robust customer support systems. Other notable trends include the development of streaming RAG approaches, which enable real-time knowledge base updates, and the use of language model augmented retrievers, which can improve domain-specific knowledge indexing. Noteworthy papers in this area include 'Query-Aware Graph Neural Networks for Enhanced Retrieval-Augmented Generation' and 'LMAR: Language Model Augmented Retriever for Domain-specific Knowledge Indexing', which demonstrate significant improvements in retrieval accuracy and efficiency.
Advancements in Retrieval-Augmented Generation
Sources
Query-Aware Graph Neural Networks for Enhanced Retrieval-Augmented Generation
Enhancing Retrieval-Augmented Generation for Electric Power Industry Customer Support
From Static to Dynamic: A Streaming RAG Approach to Real-time Knowledge Base
LMAR: Language Model Augmented Retriever for Domain-specific Knowledge Indexing
Planning Agents on an Ego-Trip: Leveraging Hybrid Ego-Graph Ensembles for Improved Tool Retrieval in Enterprise Task Planning
You Don't Need Pre-built Graphs for RAG: Retrieval Augmented Generation with Adaptive Reasoning Structures
Vec2Summ: Text Summarization via Probabilistic Sentence Embeddings
DIVER: A Multi-Stage Approach for Reasoning-intensive Information Retrieval
What Breaks Knowledge Graph based RAG? Empirical Insights into Reasoning under Incomplete Knowledge
Privacy-protected Retrieval-Augmented Generation for Knowledge Graph Question Answering
Towards Self-cognitive Exploration: Metacognitive Knowledge Graph Retrieval Augmented Generation
From Ranking to Selection: A Simple but Efficient Dynamic Passage Selector for Retrieval Augmented Generation
CS-Agent: LLM-based Community Search via Dual-agent Collaboration
AmbiGraph-Eval: Can LLMs Effectively Handle Ambiguous Graph Queries?
RAGulating Compliance: A Multi-Agent Knowledge Graph for Regulatory QA
Guided Navigation in Knowledge-Dense Environments: Structured Semantic Exploration with Guidance Graphs
Large Language Models for Summarizing Czech Historical Documents and Beyond
LeanRAG: Knowledge-Graph-Based Generation with Semantic Aggregation and Hierarchical Retrieval
FIRESPARQL: A LLM-based Framework for SPARQL Query Generation over Scholarly Knowledge Graphs