KX-02 · Analysis
Source-Grounded Research Ontology
Build a shared vocabulary and relationship map across a programme's evidence base.
Document-SupportedData-Supported
Frontier programmes accumulate incompatible terminology across teams and years. The ontology module fixes entities, quantities, conditions, and relationships to specific sources so that later reasoning is traceable.
Institute module developed within the KRYOS v6 framework ecosystem. Established scientific methods retain their conventional attribution.
When to use
When the same phenomenon is described in several vocabularies across teams or eras.
When not to use
Before intake has produced a provenance-tracked register.
Required inputs
- Evidence register from KX-01
- Programme glossary if one exists
Optional inputs
- Instrument taxonomy
- Prior ontology or data dictionary
Assumptions and constraints
- Terminology conflicts are recorded, not silently harmonised.
Process steps
- 01Extract entities, quantities, conditions, and apparatus references.
- 02Cluster synonyms and record contested definitions.
- 03Link every relationship to its supporting claim IDs.
- 04Flag relationships that rest on interpretation only.
- 05Publish the ontology with unresolved-term list.
Human review gates
- PI ratifies contested definitions before downstream use.
Primary output — Research ontology with source-linked relationships and contested-term register.
- entity_id, canonical_term, synonyms[]
- relationship_id, subject, predicate, object, supporting_claims[]
- contested_terms[] with competing definitions
Failure signals
- Relationships without supporting claim IDs.
- Silent merging of terms that describe different measurements.
Quarantine triggers
- Ontology used to imply relationships no source states.
Mini prompt
From the registered claims in [INPUT DOCUMENTS], build an ontology of entities, quantities, and conditions. Link every relationship to claim IDs. List contested terms without resolving them.
