Elamur Groupantsontology-core-selection
The groupant beside models/ontology-core-selection.groupant in the Mova repository: what the code computes, what it never does, how to go back from its outputs, its invariants, cost and checks.
Facts
Implementation material. A Mova code model describes this code at one source revision: partial models, finite checks and their stated scope. It is not a reviewed general exposition, and its labels are shown as written, never raised.
Components
| component | label | where |
|---|---|---|
| approximant | partial | forward |
| ast | not materialized | |
| composition | partial | composition |
| cost | symbolic source derived | cost |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | see source local evidence | checks |
Obernants
The typed relations around the forward model: the behaviour complement (what the code never does), the reverse reconstruction (every input compatible with an output) and specialised views.
| id | type | label | where |
|---|---|---|---|
behavior-complement | behavior_complement | partial source derived model | behavior-complement |
reverse-reconstruction | reverse_reconstruction | partial source derived model | reverse-reconstruction |
categorical-state-transport | categorical_state_transport | partial source derived model | categorical-state-transport |
sense-identity | representation_transport | partial source derived model | sense-identity |
csr-join | csr_transport | partial source derived model | csr-join |
teacher-effect | effect_boundary | partial source derived model | teacher-effect |
Model text
descriptant
ontology-core-selection: wave29
coresel uses wdwalk core resolution and corevmm key/verdict readers, build.rs consumes coregate then core.rs includes generated arrays. Graph model feeds store/Kb/query; core checks layer0 Kb properties and ancestry. Builder consumes Zip; judge/corevmm call wdvmm only in unexecuted effectful paths. Passing parser is not successful numeric code emission or teacher correctness.
This source-bound partial group organizes existing Mova code. Native evidence and excluded effects are recorded in source sidecars and the root wave29 report. Registration is not a universal proof or whole-system runtime claim.
model
ontology-core-selection
forward
Candidate rows and verdict keys → coresel applicability/rank filtering → explicit 20-column core cells → coregate categorical completeness → build-script CoreId/CORE → core Field and graph contradiction queries. Separately Builder XML/ZIP ingestion → sorted Graph SoA/CSR → store/Kb callers. Judge/VMM pure parsers map responses into bounded categorical observations; live teacher transport not executed.
invariant
Core order preserved into enum/static indices; categorical unknown/n/a/known-empty stay distinct. Valid representable graph construction deduplicates edges and preserves incoming/outgoing transpose. Core lookup requires unique OEWN key/POS to agree with Graph linear lookup; public malformed Graph state outside checked domain.
cost
Core gate O(RC) membership plus maps, selection sorts accepted numeric rows O(sum A log A); graph assembly O(W log W+E log E+L log L+sum s_w²). Text/CSR retained in memory; external teacher/I/O cost not measured.
composition
coresel uses wdwalk core resolution and corevmm key/verdict readers, build.rs consumes coregate then core.rs includes generated arrays. Graph model feeds store/Kb/query; core checks layer0 Kb properties and ancestry. Builder consumes Zip; judge/corevmm call wdvmm only in unexecuted effectful paths. Passing parser is not successful numeric code emission or teacher correctness.
domain
Synthetic UTF8 helper inputs and valid bounded CSR. No teacher/held-out/DB execution. Initial query allowed ambient sense-prior lookup; corrected four isolated runs use absent ONTO_PRIOR.
limitations
Resources available; u32 offsets and enum widths representable, CSR externally validated. Live writes, worker races, write-error handling and full data ingestion are modeled but not run.
checks
- native-commands.json: 24 cargo named tests, 2 build-script named tests, twice debug/release
- controls.json: 8 compiled controls with nonzero failure, restoration and positive reruns
- All nine syntax projections generated twice identically
- prior-isolation.json: affected query twice debug/release with absent ONTO_PRIOR
catalog
behavior-complement
Explicit no candidates cannot enter core; absent fields cannot pass completeness; wrong synset candidates cannot substitute senses. Gate success alone does not exclude invalid numeric/identifier emission, and pure response parse cannot exclude teacher or append failures.
reverse-reconstruction
Compiled core admits all rejected/lower-ranked candidates, comments, duplicate dropped values and numeric spellings preserving emitted records. Graph admits repeated edges and senses removed by assembly. Parsed risk empty admits none and malformed valid-ID rows, so response reconstruction is not unique.
categorical-state-transport
TSV known/unknown/n/a transports to Field Known/Unknown/NotApplicable; risk none to Known(empty), never Unknown. coregate validates complete categorical cells but not field-value grammar.
sense-identity
coresel filters exact synset; q_synsets accepts noun accepted/also mapping; Core SynMap collect last duplicate vs Graph find first means unique key/POS is necessary for agreement.
csr-join
Builder dedups(s,r,t) and builds both orientations; model borrowing assumes valid offsets. store validation belongs between loaded bytes and Kb queries; builder does not itself exclude cycles.
teacher-effect
corevmm/judge prompts and pure parsers can be tested without wdvmm::ask. Output append errors and in-flight worker effects remain outside demonstrated contracts; probability estimates are not measured frequencies.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/models/ontology-core-selection.groupant/