REBot: From RAG to CatRAG with Semantic Enrichment and Graph Routing
Researchers have developed REBot, an AI-powered chatbot designed to assist with academic regulation advising. The system utilizes a novel framework called CatRAG, which combines retrieval-augmented generation with graph-based reasoning. This approach integrates dense retrieval and knowledge graph reasoning, enhanced by semantic features and a category-labeled knowledge graph. An intent classifier directs queries to specific retrieval modules, ensuring accuracy and depth, and has demonstrated state-of-the-art performance on classification and question-answering tasks. AI
IMPACT This research introduces a novel framework for integrating retrieval and reasoning, potentially improving the accuracy and depth of AI-powered advisory systems.