Principles, the Minting Rule, and Diagnostics
Pre-review baseline — 2026-10-13. The inventory now reflects
prod, inspected read-only. This page preserves useful modeling questions while withdrawing outdated claims that the former schema tree and diagnostic cascade describe an already complete, validated implementation. It does not approve a replacement theory or new database changes.
Principles for reading the current record#
Distinguish fact, interpretation and proposal#
A stored frame edge, local FE, explicit constraint or structure row is implementation evidence. A definition supplies an interpretation that can be outdated. A proposed analysis or new schema is a separate decision. When these disagree, record the disagreement rather than silently making one source conform to another.
Meaning, not syntax or names#
Use the frame definition, FE meanings, relations and representative readings. Neither an uppercase label, grammatical voice, POS, number of overt arguments nor the string “movement” determines a schema or namespace. CLOSED_PATH_MOVEMENT currently has a path-shaped cast; that fact must remain visible even though its name and definition suggest a movement event.
Keep modeling axes separate#
Schema grounding, lexical profiling, namespace classification and domain membership answer different questions. Multiple proposed conceptual contributions do not give one accepted frame multiple namespaces. A transformation is not a bare-occurrence Phenomenon merely because it occurs; see ontological classification.
Inspect mappings instead of assuming inheritance#
A frame relation is not a completed FE alignment. The snapshot has 42 internal inheritance edges, of which 21 have no attached inheritance FE mappings. Composition is separately represented by 11 rel_sub_frame edges and 17 mappings. A missing row is missing evidence, not permission to synthesize it from matching labels.
Do not infer force everywhere#
Current LINK, EVENT and CHANGE do not require the old universal force analysis. FORCE_APPLICATION explicitly records force transfer. The STATIVE definition's active-maintenance/force-neutral language needs review against its actual cast and parentage, not a forced reinterpretation of all states as equilibrium.
Separate local FEs from reusable Elements#
A schema's local role inventory is not proof of canonical Element identity. All 168 selected local FEs have null idElement in this production baseline. The profiling mechanism describes repository behavior, not a completed migration of this inventory.
The historical minting rule: questions to reconsider#
Earlier documentation used six tests to distinguish a new schema from a profile of an existing base. Some current definitions still cite their numbers. The tests are retained here as design history and review prompts, not automatic authorization to create objects.
First identify the actual base and the intended kind of object: schema, reusable relation component, lexical frame, or profile. Then ask:
| Historical test | Review question | Current evidence / limitation |
|---|---|---|
| 1. Inventory | Does the candidate add a role or relation absent from the base? | OPPOSITION and CENTER_PERIPHERY each add Center over the current center-less AXIS. Their separate identities are stored facts; whether to retain both is a future design question. |
| 2. Constraint | Does it impose a constitutive restriction? | SUCCESSION's poles are explicitly constrained to EVENT. A sentient or force-related reading merely mentioned in prose is not an implemented restriction. |
| 3. Process | Does it require internal progression, not just duration? | PROCESS.Ground is explicitly constrained to SUCCESSION. The old generic AXIS account is no longer the recorded restriction. |
| 4. Invariant | Does it fix something the base leaves free? | CYCLE and CLOSED_PATH_MOVEMENT have explicit pole-equality rows. The former STATIVE-over-CHANGE/Income=Outcome example is obsolete. |
| 5. Reference | Does another independently justified object need this identity? | CHANGE.Ground has a direct constraint to DIMENSION. Such references are evidence to inspect, not proof that every referenced object must survive unchanged. |
| 6. Reusable naming | Is a named role reusable despite sharing its base's shape? | SOURCE, GOAL and VIA each have Region and a VALUE parent. Their definitions invoke reusable naming; no automatic minting decision follows. |
The former statement that every object always walks back to REGION or LINK needs a scope qualification. It holds for the current 44 schema-prefixed inheritance subgraph, not the entire Structural namespace: Boulder inherits from lexical Event. The current tree is maintained only in Schemas, rather than duplicated here with obsolete branches.
Structural similarity also does not authorize a merge. Attributes remains an Attribute frame linked by inheritance from LINK. Equating a lexical frame and a schema requires a separate review of identity, lexicalization, roles, constraints and relations; it is not established by cast size alone.
Diagnostics#
The old cascade used sentience, displacement, force, role count and converse predicates as decisive gates. It is not a safe current classifier: it routed to removed schemas, assumed the old STATIVE cast and treated argument count/converse wording as sufficient evidence.
Use the following review sequence, with an explicit defer outcome:
- Identify the task. Namespace assignment, schema selection, FE correspondence, local structure and LU profiling are separate tasks. Use the namespace procedure for namespace questions.
- Read the complete defined contribution. Is it a referent, configuration, relation, occurrence, transformation, controlled action, force transfer or combination? Psychological content can determine a namespace without requiring a same-named Structural schema.
- Inspect candidate records. Verify frame IDs, current FEs and coreness, direct parents, components, local assertions and explicit constraints. Do not take stale definitions as the current cast.
- Propose semantic correspondences. Match role meanings, not their names. Preserve the distinction between a named position (SOURCE/GOAL/VIA), a path (PATH family) and motion (MOTION family).
- Check composition separately. Controlled locomotion or effector action may involve several component structures. Inspect the actual composite-to-component mappings rather than inventing inheritance parents.
- Check what is missing. Unmapped roles, absent constraints, stale dynamic programs and contradictory definitions must remain visible. Distinguish an unrecorded analysis from an analysis shown to be incompatible.
- Compare alternatives and defer if necessary. A proposed structure must explain the lexical evidence better than the alternatives. A single grammatical test or convenient example cannot settle a mixed frame.
Useful contrasts, not automatic gates#
| Contrast | Inspect |
|---|---|
| Plain property vs maintained/resultant situation | What the whole frame entails, not a causal history supplied by world knowledge. Attribute/State namespaces do not mechanically select one schema. |
| Psychological experience vs physical action | Constitutive mental/sensory content and participant perspective; follow current namespace boundaries, including acting-causer distinctions. |
| Bare occurrence vs transformation | Whether change in position/condition is part of the definition; EVENT has one role, CHANGE has transition structure. |
| AXIS vs OPPOSITION / CENTER_PERIPHERY | Whether a genuine Center is represented and what the structure means. |
| PATH vs MOVEMENT_ALONG_A_PATH | A route/configuration versus an occurrence with a moving Protagonist and Ground constrained to PATH. |
| CHANGE vs PROCESS | Constitutive intermediate progression, not duration alone; inspect the SUCCESSION restriction. |
| MOTOR_CONTROL vs FORCE_APPLICATION | Controlled effector action versus actual transfer of force to ActedUpon. |
| Relation vs attribution | Connection among relata versus a bearer's property; FE count and converse wording only corroborate an analysis. |
The old converse-predicate table should not be used as a direct lookup: pertencer, conter or estar perto can occur in different readings and constructions. Determine the target's lexical contribution before selecting a schema or mediator.
What to record for the redesign#
For each proposal, record:
- Current frame/FE/relation IDs and the baseline timestamp.
- The current stored cast, edge direction and explicit restrictions.
- The intended interpretation and the exact conflicting evidence, if any.
- Candidate changes and alternatives, including a no-change option.
- Effects on inheritance mappings, composition, local structure, constraints, dynamic programs and lexical uses.
- Separate status for documentation clarification, proposed data change, approval and application.
Start with the C01–C12 review register. Nothing in this refresh authorizes a database correction, LU move, automatic merge or new schema.