Synthetic Community Knowledge Graphs
KinGraph
A toolkit for generating synthetic human community datasets with rich, temporally-consistent knowledge graph structures for graph ML and database pipelines.
KinGraph simulates communities of thousands of people — families, marriages, schools, employers, banks, and friendships — enforcing temporal consistency so children are never born before their parents or a marriage. Each person's simulated life is rendered as a natural-language biography, which an LLM pipeline then converts back into structured subject-relation-object triples, ready to load into Neo4j or feed graph machine learning pipelines. The result is a deterministic, reproducible testbed for knowledge graph extraction and reasoning research.
What we are exploring
The system is shaped by questions we keep returning to in our research notes. Where answers are speculative, the design is conservative; where the answers are mature, we ship against them.
Why it matters
Projects exist to be measured against outcomes, not against a launch narrative. The studio reviews each project against the standard a regulated enterprise would apply to any operational system.