Knowledge Graph
By Seme Research Team · Updated May 22, 2026
Definition
A Knowledge Graph is a network representation of entities (people, organizations, locations) and the relationships between them, stored as a graph of nodes and edges. In identity investigation, knowledge graphs visualize connections discovered during research, helping analysts understand complex networks of associations. Each node represents an entity with attributes (name, type, properties), and each edge represents a relationship (works_at, knows, published_with, located_in). Knowledge graphs enable graph-based queries such as "find all people who worked at both Company A and Company B" or "identify the shortest connection path between Person X and Person Y." Modern investigation platforms use knowledge graphs with thousands of nodes to map organizational structures, influence networks, and entity relationships.
How It Works
Knowledge graph construction in identity investigation follows four steps. Step 1 — Entity Extraction: NLP models identify entities (persons, organizations, locations, dates) from collected evidence and assign unique identifiers. Step 2 — Relationship Detection: the system identifies relationships between entities based on context (e.g., "worked at" from employment records, "co-authored" from paper listings). Step 3 — Graph Population: entities become nodes and relationships become edges, with attributes attached to both. Step 4 — Graph Analysis: algorithms such as community detection (finding clusters), centrality analysis (identifying key influencers), and path finding (discovering connection routes) are applied. The graph is continuously updated as new evidence is collected during the investigation.
Example
Building a knowledge graph for a tech company investigation reveals: Node "Alice" — works_at "TechCorp" — manages → "Bob" and "Charlie". Node "Bob" — previously_worked_at "StartupX" — co_founder_with → "Diana". Node "Diana" — board_member_at "InvestmentFund" — invested_in → "TechCorp". The graph reveals that Diana's investment fund funded TechCorp where Bob (her co-founder) now works — a previously undisclosed relationship that represents a potential conflict of interest.
Applications
- •Organizational structure mapping for corporate investigations
- •Influence network analysis in political and business contexts
- •Fraud ring detection through connection pattern analysis
- •Academic collaboration network visualization
Key Statistics
| Metric | Value | Source |
|---|---|---|
| Average Nodes per Investigation | 50-200 | Seme platform data |
| Relationship Types | 15+ | Schema design |
| Graph Algorithms Used | 6 (centrality, community, path) | Technical architecture |
| Update Frequency | Real-time during investigation | Seme architecture |