GraphRAG - Why, When and How?

Knowledge-Enhanced AI Systems

# Why GraphRAG? (Intro)
## 1.1 State of the art
State of the art alternatives - strengths and limitations of LLMs, Vectors, FTS, LPG, etc.

Demo: Comparison of different retrieval outcomes, using a real-world dataset.
Highlighting where each approach excels and fails.

## 1.2 How you index vs. What you index
It's not your RAG variety, it's your metadata

Demo: Showcase typical deficiencies

## 1.3 The Realization: Semantic Layers and Backbone

The semantic backbone: A coherent knowledge layer.

Demo: How AI can accelerate the SemLayer related tasks

# How to Implement GraphRAG (Workshop)
## 2.1 Implementation Methodology
Stages, tasks, and challenges:
- Requirements: Competency Questions Elicitation.
- Modeling automation
- Knowledge Extraction
- Enrichment and Validation
- Integration and Deployment
- Monitoring and Iteration

## 2.2 Focus on Quality
The risk of accelerated automation: Maintaining output quality

Quality pillars: Completeness, Accuracy, Consistency, Provenance

Demo: Automated mapping workflow showing:
- Dataset ingestion
- initial extraction
- Validation checks
- Competency Question Coverage audit
- Human-in-the-loop

## 2.3 Graphwise Platform Overview
Demo: GraphRAG with Graphwise Platform:
- Core Components: GraphDB, KM, AI Platform
- Functional: Sample workflows breakdown and review.
- Non-Functional: Scalability, Performance, Security, Observability
- Sample Applications