Entity Interpretation Graph — evidence into selected context [Concept]

# Entity Interpretation Graph — evidence into selected context `[Concept]`

The Entity Interpretation Graph (EIG) is an entity-scoped, derived explanation over canonical evidence. It
reuses the graph family's typed ontology and the Phase graph's immutable snapshot lifecycle, but its
terminal is a field of tendencies rather than a plan or action.

```mermaid
flowchart TB
    subgraph Evidence["canonical evidence · authoritative"]
        O["outline + Scenes<br/>actions · expressions · consequences"]
        C["entity continuity<br/>knowledge · Frames · capabilities"]
        F["formation history<br/>culture · education · practice · relationships"]
    end

    O -->|"as-of projection + stable refs"| Profile["Entity Evidence Profile<br/>coverage · perspective · evidence signature"]
    C --> Profile
    F --> Profile
    Profile -->|"coverage gate"| Generate["author or generate EIG<br/>strict JSON + reference validation"]

    subgraph EIG["Entity Interpretation Graph · derived snapshot"]
        I["indicators + formation"]
        L["dispositions · schemas · values<br/>sensitivities · inhibitions · capacities"]
        M["motives"]
        T["tendencies"]
        I -->|"supports · contradicts · shapes"| L
        L -->|"activates · amplifies · inhibits"| M
        L -->|"enables · motivates · expresses"| T
        M -->|"motivates · inhibits"| T
    end

    Generate --> I
    T --> Bind["attributable branch + moment binding"]
    Bind --> Context["effective entity context<br/>ordered reasons replace raw continuity"]
    Context --> Activate["Behaviour Reading<br/>(described extension, not built)<br/>+ situation · seat knowledge · Frames · Control · current state"]
    Activate --> Field["action field<br/>competing pulls · inhibitions · uncertainty"]
    Field --> Will["Will chooses an attributable action"]
    Will --> Consequence["accepted consequence"]
    Consequence -.->|"new evidence may make snapshot stale"| Profile
```

**Invariants**

- Evidence remains canonical; the EIG is a versioned inference at a declared branch, cursor, and perspective.
- Each snapshot is directed and may be cyclic or acyclic. Reciprocal, reinforcing, and inhibiting loops
  are legitimate when every participating reason is evidence-grounded; consequence can also become new
  evidence for a later snapshot.
- Every inferred node has valid evidence roots plus `considered`, `breaks`, and `opens`.
- The graph ends in conditional tendencies. It never issues a tool call or commits Will.
- Sparse evidence may yield `insufficient_evidence`; generation is not entitled to invent a psyche.

**Where it is specified:** [Entity Interpretation](../../specs/latent-model/README.md). Siblings:
[Phase graph](phase-graph.md) · [Reasoning graph engine](../flows/reasoning-graph-engine.md).

Raw Markdown source