Narrative Extraction with GenAI: Human-centric Knowledge Graph Creation from Text
summary
Proposes a narrative-first extraction pipeline that runs a GenAI step to summarize a document into main actors, events, and dynamics, then performs entity and relation extraction on that narrative (one-shot/custom prompts) into Neo4j/Aura. Shows a concrete reduction of noise: two witness statements shrink an analysis window from 9 hours to a 5-minute interval by isolating incident-centric times. Stores LLM outputs in a metagraph staging area for human validation with role-based access control, then merges approved nodes into the final knowledge graph. Results improve relevance and analyst confidence.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:30Unstructured data challenges
- 3:17Entity extraction with GenAI
- 5:02Narrative extraction approach
- 7:10Knowledge graph comparison
- 9:16Temporal analysis benefits
- 11:02Pipeline recap
- 12:26Human validation process
- 14:35Key benefits summary
related talks
08mHarness Graph Power With Agentic AI: An Insights Discovery Use Case
30mHow to Build the Next Wave of Graph-Based Agentic AI
01h 26mThinking in Graphs: The Cognitive Foundations of Agentic AI
21mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
27mGraph-Powered Storyworlds: Using Neo4j To Keep 1M+ Word LitRPG Epics Coherent w/ AI
19mWisdom-Driven Knowledge Augmented Generation at Scale
12mAgentic GraphRAG: Autonomous Knowledge Graph Construction and Adaptive Retrieval
30mFusing NLP and Graph: Building a Conversational Agent for Enriched Financial Data