Elamur Groupantskg-main-ingestion
The groupant beside kg/src/main.rs 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 | approximant |
| ast | partial | ../syntax.json |
| composition | partial | composition |
| cost | partial | cost |
| descriptant | present | descriptant |
| invariant | partial | invariant |
| tests | executed bounded checks | tests |
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 |
grammar_join | grammar_join | partial source derived model | grammar_join |
heap_roles | heap_roles | partial source derived model | heap_roles |
training_split | training_split | partial source derived model | training_split |
write_order | write_order | partial source derived model | write_order |
Model text
descriptant
kg-main-ingestion
Unexecuted corpus ingestion and encoded Fact relationships.
Modeling epoch — 2026-10-31
Source read before tests and models; retained prospective notes accompany the parent. Current evidence uses bounded invented inputs only. This is partial modeling, not proof or corpus accuracy.
approximant
Forward relation
Read sentences/Morph, infer names, parse/check, allocate provenance/member IDs, sort/write columns and registry, fit sid-modulo training order/frequencies then emit generation comparisons/reports.
Executable approximant gap
Exact categorical structure, identifiers, parser decisions and effects have no useful percentage metric established here. No executable surrogate, automatic substitution or 1/2/5/10% guarantee is claimed.
invariant
Invariant and domain
Each retained Fact has one provenance edge and one edge/member; CHECKED only for render success. No execution evidence for corpus quality.
Cost and resources
Retain sentences, morphology,Facts,edges; parser sum+ElogE+compression+per-fact DP. No size cap.
cost
Cost and resources
Retain sentences, morphology,Facts,edges; parser sum+ElogE+compression+per-fact DP. No size cap.
composition
Composition
Unexecuted corpus ingestion and encoded Fact relationships.
Read sentences/Morph, infer names, parse/check, allocate provenance/member IDs, sort/write columns and registry, fit sid-modulo training order/frequencies then emit generation comparisons/reports.
The parent invokes scoped transformations through the actual functions named in their source ranges. Children partition useful observations, not a complete semantic proof; boundaries can omit helpers, tests and effects.
models
behavior_complement
Rejects excluded from graph; render failures are still encoded without CHECKED. names file creation before Registry means missing store fails early.
reverse_reconstruction
Encoded member edge lacks word index and order; provenance sid recovers external reference, not original text without sentences.tsv.
grammar_join
All outgoing edges gs=f.gram; member gd=member.gram, provenance gd0.
heap_roles
Service rel, global word:uk/sentence:tatoeba/fact:tatoeba:uk heaps have separate identities.
training_split
sid%10!=0 order/frequency fitting; names derived from all inputs. Held-out loops not invoked.
write_order
Samples/text files precede edges.write and reg.save; no cross-file atomicity.
tests
Evidence
Parent source checks (a fragment inherits only the behavior they actually exercise):
- wave29::synthetic_reader_parse_and_sub_precedence
Selected kg tests: 15 named outcomes, twice debug/twice release, equal sorted outcomes. Standalone actual global/build.rs: 3 named outcomes per run, twice debug/twice optimized; build_child is normally no-op, parent fixture invokes it against isolated synthetic seeds. Eight compiled source mutations (one per source) each compiled successfully and exited101 on the focused assertion; exact source restoration followed by positive rerun.
This fragment's untested branches remain source-derived. No live graph, DB, corpus, teacher, GPU or held-out evaluation was used. Command specs and numeric exits are retained in the external wave29 graphlang receipt. Strict kg Clippy reports existing source lints; see receipt.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/kg/src/main.rs.groupant/ingestion.groupant/