Knowledge Graph Engine

Knowledge Graph Engine

AgentWiki features a dual-layer knowledge graph combining explicit human links with implicit semantic relationships and AI-extracted entity connections.

The 6 Typed Relationships

1. relates-to: General associative relationship between concepts.
2. depends-on: Strict dependency or prerequisite requirement.
3. extends: Conceptual expansion or inheritance.
4. references: Citation or supporting evidence.
5. contradicts: Opposing viewpoint or superseded decision.
6. implements: Concrete implementation of an abstract architecture or RFC.

Graph Algorithms

• BFS Traversal: N-hop neighborhood discovery (depth 1 to 3).
• Shortest Path: Dijkstra / BFS pathfinding between any two knowledge nodes.
• Ego Networks: Focused subgraphs centered on a specific document.
• Graph AI Extraction: Automated entity and relationship extraction using Workers AI models (GLM-5.2, Kimi K2.7 Code, Gemma 4).