The science behind semantic data infrastructure
We're researchers building the foundations of ontology-driven data engineering. Our work spans hypergraph theory, knowledge extraction, and the systems that make semantic infrastructure practical at scale.
By researchers from Harvard, Stanford, Max Planck Institute, and more.
Traditional graphs model pairs. Hypergraphs model reality.
In traditional knowledge graphs, every relationship connects exactly two nodes. But real operational data doesn't work that way. A supply chain event might involve a supplier, a manufacturer, a logistics provider, and a customer—simultaneously.
Hypergraphs capture these n-ary relationships natively. No artificial decomposition, no information loss. Better for complex processes, supply chains, financial flows, and any domain where relationships span multiple entities.
Pairwise edges only. N-ary relationships require artificial decomposition.
Single hyperedge captures the entire relationship. Native n-ary semantics.
What we're working on
Hypergraph Theory
Why hypergraphs over traditional graphs? N-ary relationships, complex operational data, and mathematical foundations for semantic infrastructure.
Ontology Engineering
Automated ontology construction and schema mapping. Domain-specific knowledge extraction from documents AND ontology alignment for relational sources.
Entity Resolution
Zero-shot NER, coreference resolution, and entity linking across documents. Building coherent knowledge graphs from messy real-world data.
Data Contracts
Using ontologies as semantic schemas for data contracts. Enabling interoperability and governance across federated data mesh architectures.