Microframes Inside a Schema

Baseline: prod, 2026-10-13, read-only. The Structural neighborhood contains 58 top-level structure assertions and 116 attached mediator-role bindings. These are different kinds of rows, not 174 independent assertions. See the snapshot and complete relation list.

What a structure assertion records#

A microframe reifies a reusable relation with argument roles, conventionally Domain and Range. A local schema assertion selects two FEs and a mediator naming the relation between them. See the ontological microframe documentation for the distinction between a reusable component and an ordinary lexical relation.

For the captured assertions:

Top-level entityrelation row:
  relationType = rel_structure
  idEntity1 = local Domain-side FE entity
  idEntity2 = local Range-side FE entity
  idEntity3 = mediator frame entity
  idRelation = NULL

Attached binding rows:
  local FE entity → mediator FE entity
  idRelation = top-level assertion ID

Preserve endpoint order and read the mediator's roles before paraphrasing the assertion. Domain/Range are roles here, not namespace values. The two binding rows resolve the assertion's arguments; they are not additional world-level assertions. The application's mediator mechanism can also bind selected roles of an ordinary frame, so the mediator's modeling status must be checked rather than assumed from its position in idEntity3.

A link is not necessarily a force#

The current LINK definition is explicitly force-neutral: a generic association, with any force content supplied by a more specific relation. It has Relatum1 associatedWith Relatum2 (5354401).

Consequently, “every relation is a force” and “every static relation is equilibrium” are not reliable descriptions of this production inventory. Likewise, bare EVENT has no local causal assertion; CHANGE has transition and position assertions without an inherent Agent/Cause FE.

Representative explicit assertions#

These are local stored rows, not an expansion of all inherited structure. The full per-schema list is in schema records.

OPPOSITION, CENTER_PERIPHERY, POLARITY, DEICTIC and SCALE each have explicit center-ordering and region-in-dimension rows. SUCCESSION has pole ordering and region-in-dimension rows. PATH has pole ordering and region-in-dimension rows. Presence of some repeated structure does not establish a general policy that all descendants must copy every parent assertion.

What changed from the former account#

• EVENT no longer has the old Agonist–Event has_as_cause row or an Agonist FE.
• Structural AGENTIVE, UNDERGOING and EXPERIENTIAL are not current inventory members; their former force rows must not be presented as live schema facts.
• CHANGE/TRANSITION/MOTION use Protagonist, not Antagonist.
• STATIVE no longer expresses Income = Outcome: it has a single State and two explicit location/dimension rows.
• CYCLE's equality assertion is populated, despite contrary text in its stored definition.
• The exact current motion mediator is motion_from_to, not the old prose's moves_from_to.
• PART_WHOLE has one explicit local has_part assertion; the old text's collection-membership alternative is not an additional local row in this snapshot.
• Attributes is outside Structural. Its attribution-chain interpretation is not part of this bounded local-assertion audit.

Structure, inheritance and dynamics are separate evidence#

A local row records an assertion. An inheritance edge supplies a proposed specialization relation, with attached FE correspondences where populated. A component edge identifies another structure used by a composite. None alone supplies an executed temporal sequence.

For example, CHANGE has Protagonist-at-Income and Protagonist-at-Outcome rows. Reading these as before/after requires the transition interpretation or an appropriate dynamic model; they are not evidence that both locations hold simultaneously. Conversely, stale schema-program text cannot override the current cast. The captured EVENT, CHANGE and STATIVE programs mention removed roles and were not recompiled during the documentation refresh.

Mediator DUL classifications and property-versus-description analyses need their own evidence. This snapshot resolves mediator identities and incident schema rows, but does not independently verify every mediator's ontology grounding or external theoretical source.

Reuse and future minting#

Before proposing a new microframe, inspect the existing relation's definition, direction, argument meanings and restrictions. Similar names may hide distinct relations; different names may describe the same one.

• A symmetric reversal does not create a distinct relation. An asymmetric converse may have a separately modeled identity and inverse link; the former blanket ban on converse identities is not assumed.
• A narrower filler alone may be expressible by a constraint rather than a new mediator.
• A composition of existing relations needs an explicit expansion if modeled as a named composite.
• Two FEs do not automatically make an ordinary lexical relation a Microframe.

These are review questions, not authorization to mint, merge or delete records. Direct microframe lexicalization remains a separate policy question, as documented in the ontological inventory.