Adaptive Knowledge Graphs for Dynamic RAG Systems

Session Track: Knowledge Graphs

Session Time: 08:30 pm - 09:00 pm November 6

Session description

Vaibhava Lakshmi Ravideshik will showcase how adaptive knowledge graphs, powered by machine learning, dynamically evolve to enhance RAG systems. Traditional retrieval methods struggle with static knowledge representations, but by integrating graph-based reasoning with adaptive ML techniques, RAG systems can continuously learn, refine, and optimize their responses over time. This session will explore: Machine Learning-Driven Graph Adaptation—Using LLMs, entity recognition, and embeddings to auto-update graph structures Real-Time Knowledge Evolution—Graphs that ingest, reorganize, and optimize based on new structured and unstructured data Graph-Based Query Optimization—ML-powered retrieval ranking, link prediction, and relationship inference for better AI reasoning AI Feedback Loops for Continuous Learning—Enabling multiagent systems to refine knowledge graphs using reinforcement learning and user interactions. By the end of this talk, attendees will understand how to build self-improving GraphRAG architectures using Neo4j, Cypher, vector search, and adaptive ML techniques, enabling AI systems to think more contextually, retrieve more precisely, and evolve over time.