---
title: "world-ideas-scope — Mova code model"
description: "Mova code model of world/src/ideas.rs: components with their labels, obernants and the model text."
url: "https://groupants.elamur.ai/free/en/mova/world/src/ideas.rs.groupant/scope.groupant/"
---
Elamur: [Mova](https://md.mova.elamur.ai/free/en/) · [Dictionary](https://md.dictionary.elamur.ai/free/en/) · [Research](https://md.research.elamur.ai/free/en/) · **Groupants**

This page as HTML: <https://groupants.elamur.ai/free/en/mova/world/src/ideas.rs.groupant/scope.groupant/>

# world-ideas-scope

The groupant beside world/src/ideas.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

- **kind:** groupant_fragment
- **status:** partial source bound model
- **source:** [`world/src/ideas.rs`](https://github.com/ElamurAI/mova-free/blob/main/world/src/ideas.rs)
- **source SHA-256:** `550a41d86b634fa8…`
- **covers:** Normalized rarity ranking and cross-context carry (bytes 3895–5269)
- **part of:** [world-ideas](https://md.groupants.elamur.ai/free/en/mova/world/src/ideas.rs.groupant/)
- **repository folder:** [`world/src/ideas.rs.groupant/scope.groupant`](https://github.com/ElamurAI/mova-free/tree/main/world/src/ideas.rs.groupant/scope.groupant)
- **repository commit:** `855da7bd4a7b`

> **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](#approximant) |
| **ast** | partial | `../syntax.json` |
| **composition** | partial | [composition](#composition) |
| **descriptant** | present | [descriptant](#descriptant) |
| **invariant** | partial | [invariant](#invariant) |
| **tests** | bounded checks | [tests](#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](#behavior-complement) |
| `reverse_reconstruction` | reverse_reconstruction | partial source derived model | [reverse_reconstruction](#reverse-reconstruction) |
| `document_frequency` | document_frequency | partial source derived model | [document_frequency](#document-frequency) |
| `lexicographic_ranking` | lexicographic_ranking | partial source derived model | [lexicographic_ranking](#lexicographic-ranking) |
| `query_normalization_join` | query_normalization_join | partial source derived model | [query_normalization_join](#query-normalization-join) |
| `context_selection` | context_selection | partial source derived model | [context_selection](#context-selection) |

## Model text


### descriptant

#### Normalized rarity ranking and cross-context carry

W(t)=BTreeSet of nerves::norm results over alphanumeric tokens. Read all ideas I; df(w)=number of idea rows i with w in W(i.text). Score(i,q)=Σ over w in W(i.text) intersect W(q) of floor(1_000_000/max(df(w),1)). Retain score>0, sort by descending score, kind rank held=0/principle=1/derived=2/other=3, descending time, ascending text, then take k. carried_over(ctx)=read rows with kind held AND ctx nonempty AND stored ctx!=ctx, preserving file order.

2026-10-31 review epoch: replaced generic fragment wrappers with scoped transitions, concrete exclusions and compatible antecedents. The exact byte range/hash and shared parent AST are retained. The child joins parent read() persistence to nerves::norm and yields ranked Idea clones or carry candidates. read()/record()/seed_principles implementations are outside its byte scope; parent models their malformed-input and write behavior. Ranking receives file rows but does not make durability or deduplication guarantees.

### approximant

#### Normalized rarity ranking and cross-context carry

W(t)=BTreeSet of nerves::norm results over alphanumeric tokens. Read all ideas I; df(w)=number of idea rows i with w in W(i.text). Score(i,q)=Σ over w in W(i.text) intersect W(q) of floor(1_000_000/max(df(w),1)). Retain score>0, sort by descending score, kind rank held=0/principle=1/derived=2/other=3, descending time, ascending text, then take k. carried_over(ctx)=read rows with kind held AND ctx nonempty AND stored ctx!=ctx, preserving file order.

Categorical source relation; no executable percent surrogate or approximation error bound is claimed.

### invariant

Repeated tokens within one idea/query count once. No-overlap/zero-score rows disappear; k=0 returns no rows. Kind priority precedes recency. carried_over performs exact context string comparison and does not sort or deduplicate. These scoped functions read but do not record, seed or rewrite ideas.

Cost: Whole-file read/storage O(B); building distinct-word frequency map adds ordered-map work over total distinct words per row. Ranking computes set intersections then O(N log N) sort and keeps all rows until truncation. Sum uses u64 without checked overflow; frequency counts and allocations have no explicit cap.

### composition

The child joins parent read() persistence to nerves::norm and yields ranked Idea clones or carry candidates. read()/record()/seed_principles implementations are outside its byte scope; parent models their malformed-input and write behavior. Ranking receives file rows but does not make durability or deduplication guarantees.

### models

### behavior_complement

related cannot return a zero-score idea or more than k rows. Repeated apple tokens do not multiply relevance. At equal score, a principle row cannot precede held solely because it is newer. carried_over excludes non-held rows, empty contexts and exact current context; it does not mutate their evidence or time.

### reverse_reconstruction

related(apple,1) returning a held apple row is compatible with adding unrelated rows or repeating apple in the query; the unseen lower-ranked file suffix is not reconstructible. Changing case/plural punctuation can preserve W(q). carried_over(empty) admits all held rows with any nonempty context; its output alone cannot reveal filtered rows or the current context uniquely.

### document_frequency

df counts rows containing a normalized word, not total token occurrences. Duplicate idea rows do count separately, lowering per-word weight; floor division can produce zero when df exceeds 1_000_000.

### lexicographic_ranking

Comparison key is (-score,kind_rank,-time,text). Fully equal keys retain input order under stable sort; source/evidence/context do not break ties.

### query_normalization_join

Ideas words uses nerves::norm after token splitting. It therefore inherits stop filtering and suffix loss, while BTreeSet removes repeated normalized concepts. Talk's cue matcher is not a substitute.

### context_selection

carried_over compares raw strings: whitespace/case changes in a nonempty ctx can make it different. It selects only held ideas and preserves original record fields and ordering.


### tests

ideas::wave28_dialogue::isolated_tsv_update_and_rank exercises held-before-principle equal-overlap ranking, stop-only empty query and carried_over matching/different contexts on isolated TSV. The read/record/source-preservation assertions in that test belong to the parent persistence scope; the compiled record mutant is not credited to this ranking child.

Commands and repeated outcomes: [parent evidence](https://md.groupants.elamur.ai/free/en/mova/world/src/ideas.rs.groupant/#tests), [source evidence](https://md.groupants.elamur.ai/free/en/mova/world/src/ideas.rs.groupant/). No new execution was required for this model-only refinement.