IGRMPublic research instrument
Open data

India dependency infrastructure

India Exposure DNA

A versioned fingerprint of direct dependencies for a covered firm, sector, state, asset, route, port, commodity or technology—without compressing unlike exposures into a single score.

Data day
See latest published payload
Freshness
Source-by-source status
Current state: implemented synthetic conformance foundation. Every entity, source, right, edge and value below is synthetic. The signed two-vintage fixture proves deterministic construction, coverage denominators, freshness and change detection; it is not a production map of any real company, state, asset or dependency.

The governing rule

A fingerprint, not a score

Exposure DNA keeps a freight share in freight-share units, an input dependence in input-volume units and a qualitative relationship unquantified. It shows the source policy behind freshness, the declared universe behind coverage and every known gap. Nothing is averaged merely because it can be displayed together.

5
separate exposure dimensions
0
cross-unit aggregations
100%
known gaps surfaced

Signed deterministic test vector

Inspect one fingerprint across two releases

Loading the hash-bound synthetic fingerprint…

What a production fingerprint is designed to answer

QuestionPermitted answerRefused shortcut
Which direct dependencies are recorded?Typed, versioned edges and their evidence IDsAn untraceable narrative profile
How large is each dependency?Its original unit, denominator, period and uncertaintyOne universal exposure score
How current is it?Observation age against every linked source policyA generic green “live” badge
What is missing?Unknown magnitudes, stale edges, absent dimensions and universe gapsModel-imputed certainty
What changed?Added, removed, revised and unchanged records across signed releasesCausal attribution from a version difference

The refusal boundary

  • An unsigned, rights-drifted or schema-invalid canonical release.
  • A target outside the supported India exposure entity types or its effective interval.
  • An edge with inactive endpoints, an undeclared coverage basis or a future observation.
  • A rewritten dimension partition, hidden stale edge, false count or inconsistent delta.
  • A qualitative component presented with numeric certainty.
  • A non-synthetic public claim without a complete machine-validated claim bundle.

Machine contract

python -m src.exposure_dna snapshot RELEASE.json ENTITY_ID --root /bundle
python -m src.exposure_dna compare BEFORE.json AFTER.json ENTITY_ID \
  --before-root /before --after-root /after

The engine revalidates both signed releases and their pinned policy, schema and method bytes before construction. Snapshot and delta outputs are themselves canonical-hashed. The output omits names, URLs, source content and evidence text; a presentation layer may resolve labels only from a separately governed public label set.

What must happen before production

IGRM still needs declared real-world entity universes, rights-approved source acquisition, externally controlled signing, independent edge audits, coverage and staleness evaluation, and a claim-bundle renderer. Until those gates pass, this page proves an executable and adversarially tested architecture—not real exposure coverage or institutional adoption.

Open evidence standard · Sensor Fusion · Synthetic DNA JSON · Output schema