Elamur Groupantsworld-reader-lexical
The groupant beside world/src/reader.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 source model | forward |
| ast | partial | ../syntax.json |
| composition | partial source model | composition |
| cost | partial source model | cost |
| descriptant | partial source model | epoch-memory |
| invariant | partial source model | invariant |
| tests | partial source model | 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 |
|---|---|---|---|
world-reader-lexical-behavior-complement | behavior_complement | partial source model | behavior-complement |
world-reader-lexical-reverse-reconstruction | reverse_reconstruction | partial source model | reverse-reconstruction |
world-reader-lexical-boundary | constraint_relation | partial source model | boundary |
Model text
forward
Scope: qlemmas and cover. Concatenate main content, main subject, subordinate content and subordinate subject; drop stop words and byte length <2. qlemmas unions lemma expansions. cover counts each remaining occurrence whose expansions intersect sentence content-token forms/lemmas, divided by occurrence count (zero if empty).
invariant
Coverage is in [0,1]. Repeating a word changes its multiplicity, not the lemma set. Lexical expansion belongs to ans and the compiled lexicon, not this fragment.
cost
Expected O((T+W)L) hash work for T sentence tokens and W question occurrences with L expansion cost; string bytes and allocations remain relevant.
behavior-complement
For fixed expansions, an empty retained question cannot yield positive coverage, and an absent sentence match cannot add a numerator hit. This says nothing about paraphrase adequacy.
reverse-reconstruction
The lemma-set fiber contains every permutation and repetition yielding the same union. For coverage h/n, all ordered occurrence lists with h matching entries among n are compatible; original order and multiplicity are not recoverable from the scalar alone.
boundary
Non-ASCII word length is measured in bytes; lexical equivalence is whatever ans::lemmas supplies. A set-based ratio is not an interchangeable implementation.
composition
This observation feeds both state routing and span extraction; confidence and coverage are independent gates and cannot be multiplied as calibrated probabilities.
checks
coverage_occurrences_and_set_loss tests 36 count pairs, stop words and subordinate/subject inclusion. lemma_skip_removes_forms_and_aliases checks the separate candidate-word skip helper outside this fragment's source range.
epoch-memory
2026-10-30: source-derived occurrence/set distinction, finite tests; no full parser evaluation. Syntax projection is partial, semantic AST absent.
This page in Markdown → https://md.groupants.elamur.ai/free/en/mova/world/src/reader.rs.groupant/lexical.groupant/