Knowledge Graph Visualization

[mathaix]

User Story

As a manager
I want to explore extracted entities and relationships in a visual knowledge graph
So that I can understand connections and patterns across interviews

Acceptance Criteria

Graph Visualization

• Interactive graph using library (D3.js, Cytoscape.js, or vis.js)
• Nodes represent entities (System, Process, Person, etc.)
• Edges represent relationships (INTEGRATES_WITH, RESPONSIBLE_FOR, etc.)
• Node size indicates importance (mention frequency, confidence)
• Node color by entity type
• Edge style by relationship type

Graph Navigation

• Pan and zoom
• Click node to see details panel
• Double-click to expand related nodes
• Hover for quick info tooltip
• Search to find and highlight specific entities
• Reset view button

Filtering

• Filter by entity type (show only Systems, only Processes, etc.)
• Filter by relationship type
• Filter by confidence threshold
• Filter by interview source
• Filter by date range

Entity Details Panel

When clicking an entity:

• Show all attributes
• Show confidence score
• Show source interviews (with links)
• Show supporting evidence quotes
• Show related entities
• "View in Neo4j Browser" link for power users

Layout Options

• Force-directed layout (default)
• Hierarchical layout
• Circular layout
• Custom layout by entity type

Evidence Traceability

• Click entity → see evidence nodes
• Evidence nodes link to interview transcripts
• Click edge → see evidence supporting relationship
• Visual indicator: high vs low evidence support

Performance

• Graph loads in < 3 seconds for 500 nodes
• Smooth interactions (60fps)
• Progressive loading for large graphs (1000+ nodes)
• Option to hide low-confidence entities for clarity

Export Graph

• Export as PNG/SVG image
• Export as GraphML for Gephi/Cytoscape
• Export as JSON for programmatic access

Technical Notes

• Query Neo4j for project-scoped subgraph
• Use Cypher to fetch entity + relationships + evidence
• Consider graph simplification algorithms for large graphs
• Implement graph caching with TTL

// Example query
MATCH (e:Entity {project_id: $project_id})
OPTIONAL MATCH (e)-[r]->(other:Entity {project_id: $project_id})
OPTIONAL MATCH (e)-[:SUPPORTED_BY]->(ev:Evidence)
RETURN e, r, other, ev

Definition of Done

• Graph visualization working
• Filtering and search working
• Entity details panel working
• Evidence links working
• Performance requirements met
• Code reviewed and merged

Requirement ID

RS-04

---

🤖 Generated with Claude Code