Breaking

Robert McDermott’s AI-Knowledge-Graph Turns Documents into Interactive Knowledge Graphs

September 3, 2026 at 08:34 EDT

Open Source
Foundation Models

Robert McDermott’s AI-Knowledge-Graph project converts plain-text documents into interactive knowledge graphs by extracting Subject-Predicate-Object (SPO) triplets with a configurable large language model. It then standardizes entity names, infers additional relationships, and renders the result in a browser.

MAR 2025 · AI-KNOWLEDGE-GRAPH

Plain text becomes a living graph

An LLM extracts facts from documents, aligns the entities, infers connections, and makes the network explorable.

Reported graph edges

Two example runs show scale, not model performance.

105

Gemma 3:12b run

486

Gemma 3:27b run

The runs used different inputs.

From document to network

Extract SPO

→

Standardize entities

→

Infer relationships

→

Explore the graph

Practical use

Threat analysis

Surfaces links among actors, targets, methods, and infrastructure.

Critical condition

Review the density

Large inputs can require filtering and presentation-layer tuning.

Make uncertainty visible

The graph separates what the document states from what the system infers.

Solid edges
Extracted from text

Dashed edges
Inferred relationships

Human review
Still required for claims

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.