@instruments/taxonomy 0.1.0 → 0.2.0

package/dist/crosswalks/miu.d.ts

@@ -1,4 +1,6 @@
 import { MaterialFamily, Sector, Application } from '../index.js';
+import '../match-BDI6Evro.js';
+import '../taste.js';
 
 declare const MIU_MATERIAL_FAMILIES: readonly ["wood", "wood_veneer", "engineered_wood", "bamboo", "cork", "natural_stone", "terrazzo", "ceramic_porcelain", "glass", "metal", "concrete", "plaster", "clay", "brick", "paint_coating", "wallcovering", "laminate", "solid_surface", "mineral_composite", "resin", "linoleum", "vinyl", "rubber", "carpet", "rug", "textile", "leather", "acoustic_felt", "acoustic_composite", "recycled_composite", "paper_cardboard", "biobased_composite", "greenery_planting", "lighting", "hardware", "digital_media", "water_feature", "mixed_material", "color_palette_only", "unknown"];
 type MiuMaterialFamily = (typeof MIU_MATERIAL_FAMILIES)[number];

package/dist/estate/index.d.ts

@@ -0,0 +1,233 @@
+import { C as CanonicalAlternatives } from '../match-BDI6Evro.js';
+
+declare const DECOMPOSITION: readonly Row[];
+
+type Verdict = 
+/** Maps faithfully onto one or more canonical terms. Nothing is lost. */
+"decomposed"
+/** Decomposes, but the compound carries meaning the parts do not. Needs an anchor as well. */
+ | "lossy"
+/** The canon can express both readings; the source cannot be safely mapped to one. Declines. */
+ | "ambiguous"
+/** A real term, but answering a question outside the interiors aesthetic axes. */
+ | "out-of-axis"
+/** Resists decomposition. Must be carried as an anchor, not a concept. */
+ | "residue";
+/** Why an `out-of-axis` term is out of axis. */
+type OutOfAxisReason = 
+/** Describes a material, not a room. Belongs on the material half of the canon. */
+"material-property"
+/** A query-router marker with no descriptive content at all. */
+ | "query-marker"
+/** Answers an axis that already exists here (sector, space, typology). */
+ | "other-canonical-axis"
+/** A real question no axis here answers yet, and the gap is declared rather than absorbed. */
+ | "axis-not-authored";
+interface Row {
+    /** The vocabulary id from `./sources.ts`. */
+    source: string;
+    /** The source term, verbatim. */
+    term: string;
+    verdict: Verdict;
+    /** Canonical targets asserted together. `evokes(...)` marks a relation rather than a term. */
+    onto: readonly string[];
+    /** Typed disjunction when the source is known to mean one of several bundles but cannot say which. */
+    alternatives?: CanonicalAlternatives;
+    /** Canonical terms that had to be minted on an EXISTING axis to hold this source term. */
+    mintedTerm?: readonly string[];
+    /** Canonical terms that only exist because a WHOLE AXIS had to be created. */
+    mintedAxis?: readonly string[];
+    reason?: OutOfAxisReason;
+    note?: string;
+}
+
+declare const ESTATE_REPOS: readonly ["materia", "architizer", "designround", "designshop", "materialgraph", "samplize"];
+type EstateRepo = (typeof ESTATE_REPOS)[number];
+interface ConceptCell {
+    source: string;
+    repo: EstateRepo;
+    /** The source's own spelling, verbatim. This is the thing the page exists to show. */
+    term: string;
+    row: Row;
+}
+interface Concept {
+    /** The canonical head term, fully qualified. `evokes(...)` wrappers are stripped. */
+    canonical: string;
+    /** True when the canon reaches this concept by a relation rather than by assertion. */
+    viaEvokes: boolean;
+    /** Distinct repositories carrying a term that heads here. The row's sort key. */
+    reach: number;
+    /**
+     * How many repositories could plausibly have reached this concept: those carrying any vocabulary
+     * on the same axis.
+     *
+     * NOT the repository count, and the difference started mattering the moment samplize joined the
+     * census. Samplize's vocabulary is colour and nothing else, so it can never reach `style.rustic`
+     * and counting it against that row would report a disagreement where there is only a different
+     * subject. Six repositories, but only five of them have anything to say about style, and only
+     * three about colour.
+     */
+    couldReach: number;
+    cells: readonly ConceptCell[];
+    /** Rows whose crosswalk declines, and rows carried as anchors. The page marks both. */
+    declines: number;
+    anchors: number;
+    /**
+     * How many distinct WORDINGS the sources use, after folding case and separators.
+     *
+     * The number that says whether a row is a real disagreement. `reach` of 5 with one wording means
+     * five repositories writing the same word five ways, which a crosswalk fixes trivially. `reach`
+     * of 5 with five wordings means five different ideas about what to call the thing, which it does
+     * not. Measured across the whole corpus, only one of the twenty shared concepts is spelling-only,
+     * so casing turns out to be a rounding error rather than the story - but a reader looking at
+     * `rustic / Rustic / Rustic & Artisan` cannot tell that without being told.
+     */
+    wordings: number;
+}
+declare const concepts: () => readonly Concept[];
+/** One axis's worth of the matrix: its concepts, and only the repositories that speak about it. */
+interface AxisDisagreement {
+    axis: string;
+    concepts: readonly Concept[];
+    repos: readonly EstateRepo[];
+    /** Concepts on this axis that more than one repository reaches. */
+    shared: number;
+}
+/**
+ * The disagreement matrix, split one table per axis.
+ *
+ * ONE TABLE MIXING AXES IS SPARSE BY CONSTRUCTION, and the sparseness is not a finding. Each
+ * repository speaks about some axes and not others: samplize's vocabulary is colour and nothing
+ * else, so against every style concept it is a column of dashes. Rendered as one table that reads
+ * as six repositories disagreeing when it is really six repositories talking about different
+ * subjects, and it costs enough width to break the canonical identifiers across three lines.
+ *
+ * Split by axis, every table is dense and each makes its own point. Style is the argument, at five
+ * repositories and twenty-three concepts. Place is the sharpest, because every repository reaches
+ * it and every concept on it is shared - which is the axis the design review said was the gap.
+ *
+ * An axis is only worth a table if more than one repository speaks about it and there is more than
+ * a handful to show; the rest are reported as a count rather than drawn.
+ */
+declare const disagreementsByAxis: ({ minConcepts, }?: {
+    minConcepts?: number;
+}) => readonly AxisDisagreement[];
+/** The dictionaries as columns, in the order the matrix renders them. */
+declare const MATRIX_COLUMNS: readonly {
+    id: string;
+    repo: EstateRepo;
+    label: string;
+}[];
+/** The headline the disagreement page opens with, computed rather than asserted. */
+declare const agreementSummary: () => {
+    concepts: number;
+    reachedByAll: number;
+    reachedByFourOrMore: number;
+    distinctSpellings: number;
+    exactStringMatchesAcrossRepos: number;
+    /** Concepts more than one repository reaches. */
+    shared: number;
+    /** Repositories in the census. Not every one speaks about every axis. */
+    repos: number;
+    /** Of those, the ones where every source uses the same word and differs only in spelling. */
+    spellingOnly: number;
+};
+/**
+ * The share of the DesignShop corpus the two ambiguous style values account for, as a percentage.
+ *
+ * A SHARE AND NOT A COUNT, and the distinction is the whole reason this function exists rather than
+ * a consumer reading `DESIGNSHOP_SCHEME_STYLE_COUNTS` directly. The underlying figures are a third
+ * party's catalogue volumes and are not ours to publish. The claim anything downstream actually
+ * needs is that `Modern` and `Contemporary` sit at comparable volume and together account for a
+ * large slice of the corpus - which a percentage carries exactly as well and a count carries no
+ * better.
+ *
+ * Exported so no page has to hand-type the number. A figure typed by hand is a figure that drifts
+ * from its data, and one derived from counts nobody may publish is the same disclosure a layer
+ * down.
+ */
+declare const ambiguousStyleShare: () => number;
+
+interface ResidueReport {
+    /** Every term across all seven estate vocabularies. */
+    total: number;
+    /** Terms that were never aesthetic vocabulary: query markers and material properties. */
+    excluded: number;
+    /** The denominator the headline is quoted against. */
+    aesthetic: number;
+    decomposed: number;
+    lossy: number;
+    ambiguous: number;
+    /** Terms that resist decomposition entirely and can only be carried as anchors. */
+    residue: number;
+    outOfAxis: number;
+    /** `decomposed` as a share of `aesthetic`. TAX-14's threshold was 80 per cent. */
+    cleanRate: number;
+    /**
+     * The circularity guard. The axes here were authored after reading these terms, so the headline
+     * rate is partly fitted to its own inputs. This is the share that decomposes onto canon that
+     * existed BEFORE the test - vendored from materialgraph, colorscope and architizer, or already
+     * published in this package. It is the floor the design would have reached with no new authoring
+     * at all, and it is the number to argue with.
+     */
+    preExistingRate: number;
+    preExisting: number;
+    /** Source terms that needed a term minted on an axis that already existed. */
+    neededMintedTerm: number;
+    /** Source terms that needed a whole axis that did not exist. */
+    neededMintedAxis: number;
+}
+declare const residueReport: () => ResidueReport;
+declare const mintingDemand: () => {
+    onExistingAxes: readonly (readonly [string, number])[];
+    onNewAxes: readonly (readonly [string, number])[];
+};
+/** Per-source decomposition rate, over the aesthetic subset. */
+declare const rateBySource: () => readonly {
+    source: string;
+    decomposed: number;
+    terms: number;
+}[];
+/** Every row the canon could not take cleanly. The site publishes this list in full. */
+declare const unresolved: () => readonly Row[];
+/** Real questions the estate asks that no axis here answers yet. Declared, never absorbed. */
+declare const declaredGaps: () => readonly Row[];
+
+type VocabularyKind = 
+/** A closed enum a writer must choose from. */
+"enum"
+/** A keyword set matched against free text. */
+ | "keyword-set"
+/** Free-text prose used as an embedding anchor; the label is a handle, not a value. */
+ | "embedding-anchor"
+/** Uncontrolled free text captured from a third party's index. */
+ | "captured-free-text";
+interface EstateVocabulary {
+    id: string;
+    name: string;
+    repo: string;
+    file: string;
+    snapshot: string;
+    kind: VocabularyKind;
+    /** What question the vocabulary was built to answer, in the authors' own framing. */
+    purpose: string;
+    terms: readonly string[];
+}
+declare const MATERIA_SCENE_MOOD: EstateVocabulary;
+declare const MATERIA_INTENT_KEYWORDS: EstateVocabulary;
+declare const ARCHITIZER_STYLES: EstateVocabulary;
+declare const DESIGNROUND_STYLE_ANCHORS: EstateVocabulary;
+declare const DESIGNSHOP_SCHEME_STYLES: EstateVocabulary;
+/**
+ * Scheme counts over the 1,922 schemes carrying at least one product. Bands, not counts, are what
+ * gets published (see `Term.usage`); the raw figures live here because the decomposition test has
+ * to weight by usage and cannot do that from a band.
+ */
+declare const DESIGNSHOP_SCHEME_STYLE_COUNTS: Readonly<Record<string, number>>;
+declare const DESIGNSHOP_PRODUCT_STYLES: EstateVocabulary;
+declare const MATERIALGRAPH_MS_STYLE: EstateVocabulary;
+declare const MATERIALGRAPH_MS_MOOD: EstateVocabulary;
+declare const MATERIALGRAPH_MS_CHARACTER: EstateVocabulary;
+declare const ESTATE_VOCABULARIES: readonly EstateVocabulary[];
+
+export { ARCHITIZER_STYLES, type AxisDisagreement, type Concept, type ConceptCell, DECOMPOSITION, DESIGNROUND_STYLE_ANCHORS, DESIGNSHOP_PRODUCT_STYLES, DESIGNSHOP_SCHEME_STYLES, DESIGNSHOP_SCHEME_STYLE_COUNTS, ESTATE_REPOS, ESTATE_VOCABULARIES, type EstateRepo, type EstateVocabulary, MATERIALGRAPH_MS_CHARACTER, MATERIALGRAPH_MS_MOOD, MATERIALGRAPH_MS_STYLE, MATERIA_INTENT_KEYWORDS, MATERIA_SCENE_MOOD, MATRIX_COLUMNS, type OutOfAxisReason, type ResidueReport, type Row, type Verdict, type VocabularyKind, agreementSummary, ambiguousStyleShare, concepts, declaredGaps, disagreementsByAxis, mintingDemand, rateBySource, residueReport, unresolved };