SIGNAL/LAYER is a classification API. You send event and news text. We return structured tag clusters, DFH-validated conclusions, and a query-able corpus — organized across four axes of analysis. You license the output layer. Your product consumes the signal.
Signal/Layer is not a news reader, editorial tool, or consumer interface. It is a data classification layer. The buyer licenses the output: structured tag clusters, validated conclusions, and API query access to a continuously growing classified corpus.
Every event processed by the EAF runner produces a machine-readable tag set across four axes. These are the deliverable — structured, normalized, and versioned.
The Dual-Fidelity Harmonizer runs four coherence passes on every result. What reaches the licensee is a DFH-graded conclusion, not raw output — fidelity score attached.
Query the growing classified corpus by axis, tag, domain, fidelity grade, temporal range, or any combination. The corpus is the moat. Every ingested event strengthens it.
POST any text to /api/classify. Receive full 4-axis EAF output, DFH fidelity score, and editorial conclusion in under 50ms. REST or streaming.
The EAF runner classifies every input across four independent axes. Each axis produces its own tag namespace. Tags are composable. The full cluster is the record.
Classifies the discrete action, state change, or discovery at the center of the event. Namespaces: ACT (actions), STATE (conditions), DISC (discoveries).
Characterizes velocity, magnitude, novelty, disposition, and information quality. Five orthogonal namespaces: VEL · MAG · NOV · DISP · IQ.
Maps events to institutional, economic, social, technological, environmental, and geographic domains. Enables cross-domain correlation queries.
Detects narrative manipulation, institutional bias patterns, epistemic quality issues, temporal dynamics, actor networks, and systemic risk signals.
Every classification result passes through four coherence phases before it reaches the output. Contradictory tag combinations are caught and graded — not silently dropped. The DFH grade travels with the conclusion.
Axis 1 + Axis 2 coherence. Flags contradictory velocity, IQ conflicts, disposition mismatches.
Event type → domain mapping. Ensures actions have appropriate institutional or geographic anchoring.
Magnitude vs geographic scope. Flags existential claims without geo anchoring; local events with global magnitude tags.
Axis 4 internal coherence. Catches contradictory system tags, actor attribution conflicts, temporal contradictions.
You are licensing access to a classification layer, not software you run yourself. The EAF processes your sources. The output — tag clusters, conclusions, query access — is what you integrate into your product.