@mailwoman/registry 4.8.1 → 4.10.0

package/out/resolve.d.ts

@@ -16,13 +16,56 @@
  *   key). Wiring mailwoman's parser + geocoder to turn raw rows into `SourceRecord`s is the ingest
  *   layer that sits in front of this.
  */
-import { type BlockingKey, type FellegiSunterModel } from "@mailwoman/match";
+import { type BlockingKey, type FellegiSunterModel, type GBT, type TermFrequencyTable } from "@mailwoman/match";
 import type { ResolvedEntity, SourceRecord } from "./types.js";
 /**
- * The default geocode-first scoring model: name + organization + address key + great-circle
- * distance.
+ * Cheap, parse-free normalization for the address-frequency key — uppercase, collapse whitespace,
+ * drop punctuation. Used to count how many distinct entities share an address across the WHOLE
+ * corpus (computable over millions of rows without geocoding) and to look that frequency up at
+ * match time. It's the inverse-frequency signal: a crowded clinic/billing address is weak evidence
+ * of identity; a lonely address is strong. (See
+ * docs/articles/evals/2026-06-15-nppes-dedup-benchmark.md.)
  */
-export declare function buildDefaultModel(): FellegiSunterModel<SourceRecord>;
+export declare function addressFrequencyKey(raw: string): string;
+/**
+ * Options for {@link buildDefaultModel}. Each lever is default-off, so the base model is
+ * byte-stable.
+ */
+export interface DefaultModelOptions {
+    /**
+     * Corpus-wide address-frequency table (over {@link addressFrequencyKey}) — makes the address-
+     * agreement weight **inverse to how shared the address is** (a building with 50 providers makes
+     * "same address" near-worthless evidence). The table's `value` is the record's raw address
+     * string.
+     */
+    addressFrequency?: TermFrequencyTable;
+    /**
+     * **A1 (#625):** collapse the redundant address-key + great-circle-distance comparisons into ONE
+     * {@link spatialComparison spatial-agreement} signal — an exact-key tier (where
+     * `addressFrequency`, if set, rides) over distance buckets. Removes the double-count that
+     * over-merges co-located providers (an exact key match already implies distance ≈ 0).
+     */
+    collapseSpatial?: boolean;
+    /**
+     * **A3 (#625):** add a normalized-phone exact-match comparison — a shared line is strong evidence
+     * and the secondary corroborator that lets a true same-entity link survive name drift under A2.
+     */
+    usePhone?: boolean;
+    /**
+     * Extra secondary-identifier comparisons drawn from {@link SourceRecord.attributes} (e.g.
+     * `["authorizedOfficial"]`). Each becomes an `attr:<key>` comparison AND counts toward A2
+     * corroboration — a more reliable discriminator than phone where the data has one (#625).
+     */
+    discriminators?: string[];
+}
+/**
+ * The default geocode-first scoring model: name + organization + a spatial signal. The spatial
+ * signal is either two comparisons (address-key similarity + great-circle distance — the legacy
+ * default, which double-counts) or, with {@link DefaultModelOptions.collapseSpatial}, one collapsed
+ * {@link spatialComparison}. `addressFrequency` down-weights agreement on a crowded address either
+ * way.
+ */
+export declare function buildDefaultModel(opts?: DefaultModelOptions): FellegiSunterModel<SourceRecord>;
 /** The default blocking keys: a union of location, canonical address, phone, and email. */
 export declare function defaultBlockingKeys(): BlockingKey<SourceRecord>[];
 /** Options for {@link resolveEntities}. */
@@ -40,6 +83,73 @@ export interface ResolveConfig {
      * false.
      */
     trainEM?: boolean;
+    /**
+     * Address-frequency table (over {@link addressFrequencyKey}) — down-weights address agreement by
+     * how shared the address is (a crowded clinic/billing address is weak identity evidence).
+     * **Default-on (#625):** when omitted, `resolveEntities` AUTO-COMPUTES the table over the INPUT
+     * records' addresses (the right scope for a single dataset — a crowded address within the data is
+     * down-weighted). Pass your own {@link TermFrequencyTable} (e.g. a corpus-wide one) to override,
+     * or `false` to disable (the legacy bare baseline). Ignored if `model` is supplied.
+     */
+    addressFrequency?: TermFrequencyTable | false;
+    /**
+     * A1 (#625): collapse the redundant address-key + distance pair into one
+     * {@link spatialComparison}. **Default-on (true)** — the cleaner, less-over-merging spatial model.
+     * Set `false` for the legacy two-signal baseline. Ignored if `model` is supplied.
+     */
+    collapseSpatial?: boolean;
+    /**
+     * A2 (#625): require positive name OR org corroboration ({@link CORROBORATING_FIELDS}) for a link
+     * — a shared address alone cannot merge two records. Suppresses the spatial-only links that fuse
+     * distinct co-located providers. Default false.
+     */
+    requireCorroboration?: boolean;
+    /**
+     * A3 (#625): add a normalized-phone comparison to the default model — strong evidence and the
+     * secondary corroborator that keeps A2 from killing name-drift recall. Ignored if `model` is
+     * supplied.
+     */
+    usePhone?: boolean;
+    /**
+     * A4 (#625): clustering linkage. `"single"` (default) = connected components; `"average"` =
+     * average-linkage refinement that splits a component whose sub-clusters are joined only by a weak
+     * bridge — the principled over-merge fix.
+     */
+    linkage?: "single" | "average";
+    /**
+     * Extra secondary-identifier keys (from {@link SourceRecord.attributes}) to add as comparisons +
+     * corroborators — e.g. `["authorizedOfficial"]`. Ignored if `model` is supplied.
+     */
+    discriminators?: string[];
+    /**
+     * Override the Fellegi-Sunter link weight with a LEARNED score (#603). When set, a candidate
+     * pair's match weight is this function's return value (same threshold-comparable units as the FS
+     * weight) instead of {@link scorePair}'s. Default undefined (pure FS). The blocking + clustering
+     * are unchanged, so a trained scorer can be A/B'd against the FS spine on the identical pipeline.
+     * The function is responsible for its own feature computation (e.g. the agreement pattern, which
+     * is EM-independent, plus any corpus statistics it captured).
+     *
+     * INTERACTION with {@link requireCorroboration}: the two are independent and compose, but the
+     * corroboration gate is still evaluated on the Fellegi-Sunter `contributions` (NOT the learned
+     * score) — so a learned-high pair with no positive FS name/org/phone agreement is still gated
+     * out. A learned scorer is normally trained to subsume corroboration, so use ONE or the other;
+     * combining them lets the FS gate veto the learned score, which is rarely what you want.
+     */
+    scorer?: (a: SourceRecord, b: SourceRecord) => number;
+    /**
+     * **#603: the LEARNED gradient-boosted-tree scorer — DEFAULT-ON.** Omitted or `true` uses the
+     * bundled {@link DEDUP_GBT_MODEL} (trained on the NPPES NPI-truth set; beats the Fellegi-Sunter
+     * spine ~+5pp dedup F1 held-out within a state and ~+22pp on states it never trained on, cutting
+     * the co-located over-merge). `false` opts out to the pure FS spine; pass your own {@link GBT} for
+     * a custom model. The scorer is built over the SAME collapsed-spatial + address-frequency feature
+     * model as training (via the resolved {@link addressFrequency}), independent of this call's
+     * comparison config. An explicit {@link scorer} takes precedence. When the bundled model is active
+     * and you don't set {@link threshold}, its CALIBRATED link threshold
+     * ({@link DEDUP_GBT_META}.recommendedThreshold) is used — the GBT logit isn't in FS-weight units,
+     * so 0 would over-merge. The model is NPPES/US-trained; for a very different domain, A/B it or
+     * pass `false`.
+     */
+    learnedScorer?: boolean | GBT;
 }
 /** The outcome of a resolve pass. */
 export interface ResolveResult {

package/out/resolve.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"resolve.d.ts","sourceRoot":"","sources":["../resolve.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EACN,KAAK,WAAW,EAChB,KAAK,kBAAkB,EAavB,MAAM,kBAAkB,CAAA;AACzB,OAAO,KAAK,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,YAAY,CAAA;AAS9D;;;GAGG;AACH,wBAAgB,iBAAiB,IAAI,kBAAkB,CAAC,YAAY,CAAC,CAepE;AAED,2FAA2F;AAC3F,wBAAgB,mBAAmB,IAAI,WAAW,CAAC,YAAY,CAAC,EAAE,CAOjE;AAED,2CAA2C;AAC3C,MAAM,WAAW,aAAa;IAC7B,wDAAwD;IACxD,KAAK,CAAC,EAAE,kBAAkB,CAAC,YAAY,CAAC,CAAA;IACxC,wEAAwE;IACxE,YAAY,CAAC,EAAE,WAAW,CAAC,YAAY,CAAC,EAAE,CAAA;IAC1C,6FAA6F;IAC7F,SAAS,CAAC,EAAE,MAAM,CAAA;IAClB,yEAAyE;IACzE,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB;;;OAGG;IACH,OAAO,CAAC,EAAE,OAAO,CAAA;CACjB;AAED,qCAAqC;AACrC,MAAM,WAAW,aAAa;IAC7B,QAAQ,EAAE,cAAc,EAAE,CAAA;IAC1B,mDAAmD;IACnD,cAAc,EAAE,MAAM,CAAA;IACtB,yEAAyE;IACzE,aAAa,EAAE,KAAK,CAAC;QAAE,GAAG,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;CACnD;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,SAAS,YAAY,EAAE,EAAE,MAAM,GAAE,aAAkB,GAAG,aAAa,CAsC3G"}
+{"version":3,"file":"resolve.d.ts","sourceRoot":"","sources":["../resolve.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EACN,KAAK,WAAW,EAChB,KAAK,kBAAkB,EACvB,KAAK,GAAG,EAER,KAAK,kBAAkB,EAgBvB,MAAM,kBAAkB,CAAA;AAGzB,OAAO,KAAK,EAAE,cAAc,EAAE,YAAY,EAAE,MAAM,YAAY,CAAA;AAE9D;;;;;;;GAOG;AACH,wBAAgB,mBAAmB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAMvD;AAiCD;;;GAGG;AACH,MAAM,WAAW,mBAAmB;IACnC;;;;;OAKG;IACH,gBAAgB,CAAC,EAAE,kBAAkB,CAAA;IACrC;;;;;OAKG;IACH,eAAe,CAAC,EAAE,OAAO,CAAA;IACzB;;;OAGG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;;;;OAIG;IACH,cAAc,CAAC,EAAE,MAAM,EAAE,CAAA;CACzB;AAED;;;;;;GAMG;AACH,wBAAgB,iBAAiB,CAAC,IAAI,GAAE,mBAAwB,GAAG,kBAAkB,CAAC,YAAY,CAAC,CAoElG;AAED,2FAA2F;AAC3F,wBAAgB,mBAAmB,IAAI,WAAW,CAAC,YAAY,CAAC,EAAE,CAOjE;AAED,2CAA2C;AAC3C,MAAM,WAAW,aAAa;IAC7B,wDAAwD;IACxD,KAAK,CAAC,EAAE,kBAAkB,CAAC,YAAY,CAAC,CAAA;IACxC,wEAAwE;IACxE,YAAY,CAAC,EAAE,WAAW,CAAC,YAAY,CAAC,EAAE,CAAA;IAC1C,6FAA6F;IAC7F,SAAS,CAAC,EAAE,MAAM,CAAA;IAClB,yEAAyE;IACzE,YAAY,CAAC,EAAE,MAAM,CAAA;IACrB;;;OAGG;IACH,OAAO,CAAC,EAAE,OAAO,CAAA;IACjB;;;;;;;OAOG;IACH,gBAAgB,CAAC,EAAE,kBAAkB,GAAG,KAAK,CAAA;IAC7C;;;;OAIG;IACH,eAAe,CAAC,EAAE,OAAO,CAAA;IACzB;;;;OAIG;IACH,oBAAoB,CAAC,EAAE,OAAO,CAAA;IAC9B;;;;OAIG;IACH,QAAQ,CAAC,EAAE,OAAO,CAAA;IAClB;;;;OAIG;IACH,OAAO,CAAC,EAAE,QAAQ,GAAG,SAAS,CAAA;IAC9B;;;OAGG;IACH,cAAc,CAAC,EAAE,MAAM,EAAE,CAAA;IACzB;;;;;;;;;;;;;OAaG;IACH,MAAM,CAAC,EAAE,CAAC,CAAC,EAAE,YAAY,EAAE,CAAC,EAAE,YAAY,KAAK,MAAM,CAAA;IACrD;;;;;;;;;;;;OAYG;IACH,aAAa,CAAC,EAAE,OAAO,GAAG,GAAG,CAAA;CAC7B;AAED,qCAAqC;AACrC,MAAM,WAAW,aAAa;IAC7B,QAAQ,EAAE,cAAc,EAAE,CAAA;IAC1B,mDAAmD;IACnD,cAAc,EAAE,MAAM,CAAA;IACtB,yEAAyE;IACzE,aAAa,EAAE,KAAK,CAAC;QAAE,GAAG,EAAE,MAAM,CAAC;QAAC,IAAI,EAAE,MAAM,CAAA;KAAE,CAAC,CAAA;CACnD;AAED;;;GAGG;AACH,wBAAgB,eAAe,CAAC,OAAO,EAAE,SAAS,YAAY,EAAE,EAAE,MAAM,GAAE,aAAkB,GAAG,aAAa,CAiG3G"}

package/out/resolve.js

@@ -16,7 +16,24 @@
  *   key). Wiring mailwoman's parser + geocoder to turn raw rows into `SourceRecord`s is the ingest
  *   layer that sits in front of this.
  */
-import { DEFAULT_DISTANCE_LEVELS, agreementPattern, block, cluster, distanceComparison, estimateParameters, exactKey, geoCellKey, representative, scorePair, similarityComparison, } from "@mailwoman/match";
+import { DEFAULT_DISTANCE_LEVELS, DEFAULT_SPATIAL_LEVELS, agreementPattern, block, buildTermFrequencyTable, cluster, distanceComparison, estimateParameters, exactKey, geoCellKey, representative, scorePair, similarityComparison, spatialComparison, withTermFrequency, } from "@mailwoman/match";
+import { createGbtScorer } from "./learned-scorer.js";
+import { DEDUP_GBT_META, DEDUP_GBT_MODEL } from "./models/dedup-gbt-en-us.js";
+/**
+ * Cheap, parse-free normalization for the address-frequency key — uppercase, collapse whitespace,
+ * drop punctuation. Used to count how many distinct entities share an address across the WHOLE
+ * corpus (computable over millions of rows without geocoding) and to look that frequency up at
+ * match time. It's the inverse-frequency signal: a crowded clinic/billing address is weak evidence
+ * of identity; a lonely address is strong. (See
+ * docs/articles/evals/2026-06-15-nppes-dedup-benchmark.md.)
+ */
+export function addressFrequencyKey(raw) {
+    return raw
+        .toUpperCase()
+        .replace(/[^A-Z0-9]+/g, " ")
+        .trim()
+        .replace(/\s+/g, " ");
+}
 /** Default tiered levels for a name-like text field. `m`/`u` are EM-estimable seeds. */
 const NAME_LEVELS = [
     { label: "exact", minSimilarity: 1.0, m: 0.8, u: 0.01 },
@@ -24,17 +41,89 @@ const NAME_LEVELS = [
     { label: "different", minSimilarity: 0, m: 0.05, u: 0.96 },
 ];
 /**
- * The default geocode-first scoring model: name + organization + address key + great-circle
- * distance.
+ * Exact-vs-different levels for a normalized phone. A shared line is strong, rarely-coincidental
+ * evidence.
+ */
+const PHONE_LEVELS = [
+    { label: "exact", minSimilarity: 1.0, m: 0.6, u: 0.002 },
+    { label: "different", minSimilarity: 0, m: 0.4, u: 0.998 },
+];
+/** Last-10-digits normalization for phone agreement (drops country code, punctuation, extensions). */
+function normalizePhone(raw) {
+    if (!raw)
+        return null;
+    const digits = raw.replace(/\D+/g, "");
+    return digits.length >= 10 ? digits.slice(-10) : digits || null;
+}
+/**
+ * The identity-corroborating comparisons (person name, organization, phone). A2 (#625,
+ * {@link ResolveConfig.requireCorroboration}) requires at least one of these to _positively_ agree
+ * before a pair may link — a shared address alone is not identity. Phone (A3) is the secondary
+ * identifier that rescues a true same-entity link across name drift.
  */
-export function buildDefaultModel() {
+const CORROBORATING_FIELDS = new Set(["given", "family", "organization", "phone"]);
+/**
+ * The default geocode-first scoring model: name + organization + a spatial signal. The spatial
+ * signal is either two comparisons (address-key similarity + great-circle distance — the legacy
+ * default, which double-counts) or, with {@link DefaultModelOptions.collapseSpatial}, one collapsed
+ * {@link spatialComparison}. `addressFrequency` down-weights agreement on a crowded address either
+ * way.
+ */
+export function buildDefaultModel(opts = {}) {
+    const identity = [
+        similarityComparison({ name: "given", extract: (r) => r.name?.given, levels: NAME_LEVELS }),
+        similarityComparison({ name: "family", extract: (r) => r.name?.family, levels: NAME_LEVELS }),
+        similarityComparison({
+            name: "organization",
+            extract: (r) => r.organization?.canonical,
+            levels: NAME_LEVELS,
+        }),
+    ];
+    if (opts.usePhone) {
+        identity.push(similarityComparison({
+            name: "phone",
+            extract: (r) => normalizePhone(r.phone),
+            similarity: (a, b) => (a === b ? 1 : 0), // exact normalized-digit match only
+            levels: PHONE_LEVELS,
+        }));
+    }
+    for (const key of opts.discriminators ?? []) {
+        identity.push(similarityComparison({
+            name: `attr:${key}`,
+            extract: (r) => r.attributes?.[key],
+            levels: NAME_LEVELS,
+        }));
+    }
+    if (opts.collapseSpatial) {
+        let spatial = spatialComparison({
+            name: "spatial",
+            key: (r) => r.address?.canonicalKey,
+            coordinate: (r) => r.address?.geocode?.coordinate,
+            levels: DEFAULT_SPATIAL_LEVELS,
+        });
+        if (opts.addressFrequency) {
+            spatial = withTermFrequency(spatial, {
+                table: opts.addressFrequency,
+                value: (a) => a.address?.raw ?? null,
+                levels: [0], // the exact same-key tier
+            });
+        }
+        return { lambda: 0.0001, comparisons: [...identity, spatial] };
+    }
+    // Legacy two-signal spatial: address-key similarity + great-circle distance (redundant; A1 collapses it).
+    let address = similarityComparison({
+        name: "address",
+        extract: (r) => r.address?.canonicalKey,
+        levels: NAME_LEVELS,
+    });
+    if (opts.addressFrequency) {
+        address = withTermFrequency(address, { table: opts.addressFrequency, value: (a) => a.address?.raw ?? null });
+    }
     return {
         lambda: 0.0001,
         comparisons: [
-            similarityComparison({ name: "given", extract: (r) => r.name?.given, levels: NAME_LEVELS }),
-            similarityComparison({ name: "family", extract: (r) => r.name?.family, levels: NAME_LEVELS }),
-            similarityComparison({ name: "organization", extract: (r) => r.organization?.canonical, levels: NAME_LEVELS }),
-            similarityComparison({ name: "address", extract: (r) => r.address?.canonicalKey, levels: NAME_LEVELS }),
+            ...identity,
+            address,
             distanceComparison({
                 name: "distance",
                 extract: (r) => r.address?.geocode?.coordinate,
@@ -57,33 +146,88 @@ export function defaultBlockingKeys() {
  * exactly one entity (a record with no confident link is its own singleton entity).
  */
 export function resolveEntities(records, config = {}) {
-    const model = config.model ?? buildDefaultModel();
+    // The proven levers are DEFAULT-ON (#625): the address-frequency down-weight (auto-computed over the
+    // input records when not supplied; `false` disables) + the collapsed spatial signal (A1). A new
+    // caller gets the strong config out of the box; pass explicit values to override.
+    const addressFrequency = config.addressFrequency === false
+        ? undefined
+        : (config.addressFrequency ??
+            buildTermFrequencyTable(records.map((r) => r.address?.raw), { normalize: addressFrequencyKey }));
+    const collapseSpatial = config.collapseSpatial ?? true;
+    const model = config.model ??
+        buildDefaultModel({
+            addressFrequency,
+            collapseSpatial,
+            usePhone: config.usePhone,
+            discriminators: config.discriminators,
+        });
     const blockingKeys = config.blockingKeys ?? defaultBlockingKeys();
-    const threshold = config.threshold ?? 0;
+    // #603: the learned scorer is DEFAULT-ON. An explicit `scorer` overrides everything; otherwise
+    // `learnedScorer === false` opts out to the FS spine, a GBT supplies a custom model, and
+    // `true`/omitted uses the bundled DEDUP_GBT_MODEL. The scorer is built over the FIXED
+    // collapsed-spatial + address-frequency feature model (matching training, independent of this call's
+    // comparison config), using the resolved address-frequency table.
+    let scorer = config.scorer;
+    let usingBundledModel = false;
+    if (!scorer && config.learnedScorer !== false) {
+        const gbt = config.learnedScorer === undefined || config.learnedScorer === true ? DEDUP_GBT_MODEL : config.learnedScorer;
+        usingBundledModel = gbt === DEDUP_GBT_MODEL;
+        scorer = createGbtScorer({
+            model: gbt,
+            comparisons: buildDefaultModel({ collapseSpatial: true, addressFrequency }).comparisons,
+            addressFrequency: addressFrequency ?? buildTermFrequencyTable([], { normalize: addressFrequencyKey }),
+        });
+    }
+    // Threshold: an explicit value wins; else the bundled model's CALIBRATED threshold when it's active
+    // (its logit isn't in FS-weight units, so 0 would over-merge); else 0 (FS spine or a custom model).
+    const threshold = config.threshold ?? (usingBundledModel ? DEDUP_GBT_META.recommendedThreshold : 0);
     const { pairs, droppedBlocks } = block(records, blockingKeys, { maxBlockSize: config.maxBlockSize });
     let scoringModel = model;
     if (config.trainEM && pairs.length > 0) {
         const patterns = pairs.map(([a, b]) => agreementPattern(model.comparisons, a, b));
         scoringModel = estimateParameters(model, patterns).model;
     }
-    const links = pairs.map(([a, b]) => ({
-        a,
-        b,
-        weight: scorePair(scoringModel, a, b).weight,
-    }));
-    const clusters = cluster(records, links, { threshold });
+    const links = pairs.map(([a, b]) => {
+        const score = scorePair(scoringModel, a, b);
+        // #603: a learned scorer (explicit `scorer` or the opt-in `learnedScorer`) replaces the FS weight
+        // (same clustering + threshold semantics).
+        let weight = scorer ? scorer(a, b) : score.weight;
+        // A2 (#625): a link must carry positive name OR org corroboration — a shared (even down-weighted)
+        // address alone is not identity. Spatial-only pairs are suppressed below any threshold.
+        if (config.requireCorroboration) {
+            const corroborated = score.contributions.some((c) => (CORROBORATING_FIELDS.has(c.name) || c.name.startsWith("attr:")) && c.weight > 0);
+            if (!corroborated)
+                weight = Number.NEGATIVE_INFINITY;
+        }
+        return { a, b, weight };
+    });
+    const clusters = cluster(records, links, { threshold, linkage: config.linkage });
+    // Cohesion = the weakest within-cluster link weight (how tightly an entity holds together). Compute it
+    // in ONE pass over links via a record→cluster index, not by filtering every link for every cluster —
+    // the latter is O(clusters × links) and dominates the resolve at scale.
+    const clusterOf = new Map();
+    clusters.forEach((group, i) => {
+        for (const record of group)
+            clusterOf.set(record, i);
+    });
+    const minIntraWeight = new Array(clusters.length).fill(Infinity);
+    for (const link of links) {
+        if (link.weight < threshold)
+            continue;
+        const ci = clusterOf.get(link.a);
+        if (ci === undefined || ci !== clusterOf.get(link.b))
+            continue;
+        if (link.weight < minIntraWeight[ci])
+            minIntraWeight[ci] = link.weight;
+    }
     const entities = clusters.map((group, i) => {
-        const members = new Set(group);
-        const intraWeights = links
-            .filter((link) => link.weight >= threshold && members.has(link.a) && members.has(link.b))
-            .map((link) => link.weight);
         const rep = representative(group) ?? group[0];
         return {
             id: `entity-${i}`,
             records: group,
             representative: rep,
             coordinate: rep.address?.geocode?.coordinate ?? undefined,
-            cohesion: group.length > 1 && intraWeights.length > 0 ? Math.min(...intraWeights) : null,
+            cohesion: group.length > 1 && minIntraWeight[i] !== Infinity ? minIntraWeight[i] : null,
         };
     });
     return { entities, candidatePairs: pairs.length, droppedBlocks };

package/out/resolve.js.map

@@ -1 +1 @@
-{"version":3,"file":"resolve.js","sourceRoot":"","sources":["../resolve.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EAIN,uBAAuB,EACvB,gBAAgB,EAChB,KAAK,EACL,OAAO,EACP,kBAAkB,EAClB,kBAAkB,EAClB,QAAQ,EACR,UAAU,EACV,cAAc,EACd,SAAS,EACT,oBAAoB,GACpB,MAAM,kBAAkB,CAAA;AAGzB,wFAAwF;AACxF,MAAM,WAAW,GAAsB;IACtC,EAAE,KAAK,EAAE,OAAO,EAAE,aAAa,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,MAAM,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE;IACxD,EAAE,KAAK,EAAE,WAAW,EAAE,aAAa,EAAE,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE;CAC1D,CAAA;AAED;;;GAGG;AACH,MAAM,UAAU,iBAAiB;IAChC,OAAO;QACN,MAAM,EAAE,MAAM;QACd,WAAW,EAAE;YACZ,oBAAoB,CAAC,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;YAC3F,oBAAoB,CAAC,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;YAC7F,oBAAoB,CAAC,EAAE,IAAI,EAAE,cAAc,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,YAAY,EAAE,SAAS,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;YAC9G,oBAAoB,CAAC,EAAE,IAAI,EAAE,SAAS,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,YAAY,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;YACvG,kBAAkB,CAAC;gBAClB,IAAI,EAAE,UAAU;gBAChB,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU;gBAC9C,MAAM,EAAE,uBAAuB;aAC/B,CAAC;SACF;KACD,CAAA;AACF,CAAC;AAED,2FAA2F;AAC3F,MAAM,UAAU,mBAAmB;IAClC,OAAO;QACN,UAAU,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU,CAAC;QACjD,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,YAAY,CAAC;QACxC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;QACxB,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;KACxB,CAAA;AACF,CAAC;AA4BD;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,OAAgC,EAAE,SAAwB,EAAE;IAC3F,MAAM,KAAK,GAAG,MAAM,CAAC,KAAK,IAAI,iBAAiB,EAAE,CAAA;IACjD,MAAM,YAAY,GAAG,MAAM,CAAC,YAAY,IAAI,mBAAmB,EAAE,CAAA;IACjE,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,CAAC,CAAA;IAEvC,MAAM,EAAE,KAAK,EAAE,aAAa,EAAE,GAAG,KAAK,CAAC,OAAO,EAAE,YAAY,EAAE,EAAE,YAAY,EAAE,MAAM,CAAC,YAAY,EAAE,CAAC,CAAA;IAEpG,IAAI,YAAY,GAAG,KAAK,CAAA;IACxB,IAAI,MAAM,CAAC,OAAO,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,KAAK,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACjF,YAAY,GAAG,kBAAkB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAA;IACzD,CAAC;IAED,MAAM,KAAK,GAA+B,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC;QAChE,CAAC;QACD,CAAC;QACD,MAAM,EAAE,SAAS,CAAC,YAAY,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,MAAM;KAC5C,CAAC,CAAC,CAAA;IAEH,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,EAAE,KAAK,EAAE,EAAE,SAAS,EAAE,CAAC,CAAA;IAEvD,MAAM,QAAQ,GAAqB,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,CAAC,EAAE,EAAE;QAC5D,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,KAAK,CAAC,CAAA;QAC9B,MAAM,YAAY,GAAG,KAAK;aACxB,MAAM,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,IAAI,SAAS,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC;aACxF,GAAG,CAAC,CAAC,IAAI,EAAE,EAAE,CAAC,IAAI,CAAC,MAAM,CAAC,CAAA;QAC5B,MAAM,GAAG,GAAG,cAAc,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,CAAC,CAAE,CAAA;QAE9C,OAAO;YACN,EAAE,EAAE,UAAU,CAAC,EAAE;YACjB,OAAO,EAAE,KAAK;YACd,cAAc,EAAE,GAAG;YACnB,UAAU,EAAE,GAAG,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU,IAAI,SAAS;YACzD,QAAQ,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,YAAY,CAAC,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,IAAI,CAAC,GAAG,CAAC,GAAG,YAAY,CAAC,CAAC,CAAC,CAAC,IAAI;SACxF,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,KAAK,CAAC,MAAM,EAAE,aAAa,EAAE,CAAA;AACjE,CAAC"}
+{"version":3,"file":"resolve.js","sourceRoot":"","sources":["../resolve.ts"],"names":[],"mappings":"AAAA;;;;;;;;;;;;;;;;;GAiBG;AAGH,OAAO,EAMN,uBAAuB,EACvB,sBAAsB,EACtB,gBAAgB,EAChB,KAAK,EACL,uBAAuB,EACvB,OAAO,EACP,kBAAkB,EAClB,kBAAkB,EAClB,QAAQ,EACR,UAAU,EACV,cAAc,EACd,SAAS,EACT,oBAAoB,EACpB,iBAAiB,EACjB,iBAAiB,GACjB,MAAM,kBAAkB,CAAA;AACzB,OAAO,EAAE,eAAe,EAAE,MAAM,qBAAqB,CAAA;AACrD,OAAO,EAAE,cAAc,EAAE,eAAe,EAAE,MAAM,6BAA6B,CAAA;AAG7E;;;;;;;GAOG;AACH,MAAM,UAAU,mBAAmB,CAAC,GAAW;IAC9C,OAAO,GAAG;SACR,WAAW,EAAE;SACb,OAAO,CAAC,aAAa,EAAE,GAAG,CAAC;SAC3B,IAAI,EAAE;SACN,OAAO,CAAC,MAAM,EAAE,GAAG,CAAC,CAAA;AACvB,CAAC;AAED,wFAAwF;AACxF,MAAM,WAAW,GAAsB;IACtC,EAAE,KAAK,EAAE,OAAO,EAAE,aAAa,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,IAAI,EAAE;IACvD,EAAE,KAAK,EAAE,MAAM,EAAE,aAAa,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE;IACxD,EAAE,KAAK,EAAE,WAAW,EAAE,aAAa,EAAE,CAAC,EAAE,CAAC,EAAE,IAAI,EAAE,CAAC,EAAE,IAAI,EAAE;CAC1D,CAAA;AAED;;;GAGG;AACH,MAAM,YAAY,GAAsB;IACvC,EAAE,KAAK,EAAE,OAAO,EAAE,aAAa,EAAE,GAAG,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,KAAK,EAAE;IACxD,EAAE,KAAK,EAAE,WAAW,EAAE,aAAa,EAAE,CAAC,EAAE,CAAC,EAAE,GAAG,EAAE,CAAC,EAAE,KAAK,EAAE;CAC1D,CAAA;AAED,sGAAsG;AACtG,SAAS,cAAc,CAAC,GAA8B;IACrD,IAAI,CAAC,GAAG;QAAE,OAAO,IAAI,CAAA;IACrB,MAAM,MAAM,GAAG,GAAG,CAAC,OAAO,CAAC,MAAM,EAAE,EAAE,CAAC,CAAA;IACtC,OAAO,MAAM,CAAC,MAAM,IAAI,EAAE,CAAC,CAAC,CAAC,MAAM,CAAC,KAAK,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,MAAM,IAAI,IAAI,CAAA;AAChE,CAAC;AAED;;;;;GAKG;AACH,MAAM,oBAAoB,GAAG,IAAI,GAAG,CAAC,CAAC,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,OAAO,CAAC,CAAC,CAAA;AAkClF;;;;;;GAMG;AACH,MAAM,UAAU,iBAAiB,CAAC,OAA4B,EAAE;IAC/D,MAAM,QAAQ,GAAG;QAChB,oBAAoB,CAAe,EAAE,IAAI,EAAE,OAAO,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,KAAK,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;QACzG,oBAAoB,CAAe,EAAE,IAAI,EAAE,QAAQ,EAAE,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC;QAC3G,oBAAoB,CAAe;YAClC,IAAI,EAAE,cAAc;YACpB,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,YAAY,EAAE,SAAS;YACzC,MAAM,EAAE,WAAW;SACnB,CAAC;KACF,CAAA;IACD,IAAI,IAAI,CAAC,QAAQ,EAAE,CAAC;QACnB,QAAQ,CAAC,IAAI,CACZ,oBAAoB,CAAe;YAClC,IAAI,EAAE,OAAO;YACb,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,cAAc,CAAC,CAAC,CAAC,KAAK,CAAC;YACvC,UAAU,EAAE,CAAC,CAAC,EAAE,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,oCAAoC;YAC7E,MAAM,EAAE,YAAY;SACpB,CAAC,CACF,CAAA;IACF,CAAC;IACD,KAAK,MAAM,GAAG,IAAI,IAAI,CAAC,cAAc,IAAI,EAAE,EAAE,CAAC;QAC7C,QAAQ,CAAC,IAAI,CACZ,oBAAoB,CAAe;YAClC,IAAI,EAAE,QAAQ,GAAG,EAAE;YACnB,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,UAAU,EAAE,CAAC,GAAG,CAAC;YACnC,MAAM,EAAE,WAAW;SACnB,CAAC,CACF,CAAA;IACF,CAAC;IAED,IAAI,IAAI,CAAC,eAAe,EAAE,CAAC;QAC1B,IAAI,OAAO,GAAG,iBAAiB,CAAe;YAC7C,IAAI,EAAE,SAAS;YACf,GAAG,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,YAAY;YACnC,UAAU,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU;YACjD,MAAM,EAAE,sBAAsB;SAC9B,CAAC,CAAA;QACF,IAAI,IAAI,CAAC,gBAAgB,EAAE,CAAC;YAC3B,OAAO,GAAG,iBAAiB,CAAC,OAAO,EAAE;gBACpC,KAAK,EAAE,IAAI,CAAC,gBAAgB;gBAC5B,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,GAAG,IAAI,IAAI;gBACpC,MAAM,EAAE,CAAC,CAAC,CAAC,EAAE,0BAA0B;aACvC,CAAC,CAAA;QACH,CAAC;QACD,OAAO,EAAE,MAAM,EAAE,MAAM,EAAE,WAAW,EAAE,CAAC,GAAG,QAAQ,EAAE,OAAO,CAAC,EAAE,CAAA;IAC/D,CAAC;IAED,0GAA0G;IAC1G,IAAI,OAAO,GAAG,oBAAoB,CAAe;QAChD,IAAI,EAAE,SAAS;QACf,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,YAAY;QACvC,MAAM,EAAE,WAAW;KACnB,CAAC,CAAA;IACF,IAAI,IAAI,CAAC,gBAAgB,EAAE,CAAC;QAC3B,OAAO,GAAG,iBAAiB,CAAC,OAAO,EAAE,EAAE,KAAK,EAAE,IAAI,CAAC,gBAAgB,EAAE,KAAK,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,GAAG,IAAI,IAAI,EAAE,CAAC,CAAA;IAC7G,CAAC;IACD,OAAO;QACN,MAAM,EAAE,MAAM;QACd,WAAW,EAAE;YACZ,GAAG,QAAQ;YACX,OAAO;YACP,kBAAkB,CAAC;gBAClB,IAAI,EAAE,UAAU;gBAChB,OAAO,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU;gBAC9C,MAAM,EAAE,uBAAuB;aAC/B,CAAC;SACF;KACD,CAAA;AACF,CAAC;AAED,2FAA2F;AAC3F,MAAM,UAAU,mBAAmB;IAClC,OAAO;QACN,UAAU,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU,CAAC;QACjD,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,YAAY,CAAC;QACxC,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;QACxB,QAAQ,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,KAAK,CAAC;KACxB,CAAA;AACF,CAAC;AA+FD;;;GAGG;AACH,MAAM,UAAU,eAAe,CAAC,OAAgC,EAAE,SAAwB,EAAE;IAC3F,qGAAqG;IACrG,gGAAgG;IAChG,kFAAkF;IAClF,MAAM,gBAAgB,GACrB,MAAM,CAAC,gBAAgB,KAAK,KAAK;QAChC,CAAC,CAAC,SAAS;QACX,CAAC,CAAC,CAAC,MAAM,CAAC,gBAAgB;YACzB,uBAAuB,CACtB,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,OAAO,EAAE,GAAG,CAAC,EAClC,EAAE,SAAS,EAAE,mBAAmB,EAAE,CAClC,CAAC,CAAA;IACL,MAAM,eAAe,GAAG,MAAM,CAAC,eAAe,IAAI,IAAI,CAAA;IACtD,MAAM,KAAK,GACV,MAAM,CAAC,KAAK;QACZ,iBAAiB,CAAC;YACjB,gBAAgB;YAChB,eAAe;YACf,QAAQ,EAAE,MAAM,CAAC,QAAQ;YACzB,cAAc,EAAE,MAAM,CAAC,cAAc;SACrC,CAAC,CAAA;IACH,MAAM,YAAY,GAAG,MAAM,CAAC,YAAY,IAAI,mBAAmB,EAAE,CAAA;IAEjE,+FAA+F;IAC/F,yFAAyF;IACzF,sFAAsF;IACtF,qGAAqG;IACrG,kEAAkE;IAClE,IAAI,MAAM,GAAG,MAAM,CAAC,MAAM,CAAA;IAC1B,IAAI,iBAAiB,GAAG,KAAK,CAAA;IAC7B,IAAI,CAAC,MAAM,IAAI,MAAM,CAAC,aAAa,KAAK,KAAK,EAAE,CAAC;QAC/C,MAAM,GAAG,GACR,MAAM,CAAC,aAAa,KAAK,SAAS,IAAI,MAAM,CAAC,aAAa,KAAK,IAAI,CAAC,CAAC,CAAC,eAAe,CAAC,CAAC,CAAC,MAAM,CAAC,aAAa,CAAA;QAC7G,iBAAiB,GAAG,GAAG,KAAK,eAAe,CAAA;QAC3C,MAAM,GAAG,eAAe,CAAC;YACxB,KAAK,EAAE,GAAG;YACV,WAAW,EAAE,iBAAiB,CAAC,EAAE,eAAe,EAAE,IAAI,EAAE,gBAAgB,EAAE,CAAC,CAAC,WAAW;YACvF,gBAAgB,EAAE,gBAAgB,IAAI,uBAAuB,CAAC,EAAE,EAAE,EAAE,SAAS,EAAE,mBAAmB,EAAE,CAAC;SACrG,CAAC,CAAA;IACH,CAAC;IACD,oGAAoG;IACpG,oGAAoG;IACpG,MAAM,SAAS,GAAG,MAAM,CAAC,SAAS,IAAI,CAAC,iBAAiB,CAAC,CAAC,CAAC,cAAc,CAAC,oBAAoB,CAAC,CAAC,CAAC,CAAC,CAAC,CAAA;IAEnG,MAAM,EAAE,KAAK,EAAE,aAAa,EAAE,GAAG,KAAK,CAAC,OAAO,EAAE,YAAY,EAAE,EAAE,YAAY,EAAE,MAAM,CAAC,YAAY,EAAE,CAAC,CAAA;IAEpG,IAAI,YAAY,GAAG,KAAK,CAAA;IACxB,IAAI,MAAM,CAAC,OAAO,IAAI,KAAK,CAAC,MAAM,GAAG,CAAC,EAAE,CAAC;QACxC,MAAM,QAAQ,GAAG,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,gBAAgB,CAAC,KAAK,CAAC,WAAW,EAAE,CAAC,EAAE,CAAC,CAAC,CAAC,CAAA;QACjF,YAAY,GAAG,kBAAkB,CAAC,KAAK,EAAE,QAAQ,CAAC,CAAC,KAAK,CAAA;IACzD,CAAC;IAED,MAAM,KAAK,GAA+B,KAAK,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE;QAC9D,MAAM,KAAK,GAAG,SAAS,CAAC,YAAY,EAAE,CAAC,EAAE,CAAC,CAAC,CAAA;QAC3C,kGAAkG;QAClG,2CAA2C;QAC3C,IAAI,MAAM,GAAG,MAAM,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,KAAK,CAAC,MAAM,CAAA;QACjD,kGAAkG;QAClG,wFAAwF;QACxF,IAAI,MAAM,CAAC,oBAAoB,EAAE,CAAC;YACjC,MAAM,YAAY,GAAG,KAAK,CAAC,aAAa,CAAC,IAAI,CAC5C,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,oBAAoB,CAAC,GAAG,CAAC,CAAC,CAAC,IAAI,CAAC,IAAI,CAAC,CAAC,IAAI,CAAC,UAAU,CAAC,OAAO,CAAC,CAAC,IAAI,CAAC,CAAC,MAAM,GAAG,CAAC,CACvF,CAAA;YACD,IAAI,CAAC,YAAY;gBAAE,MAAM,GAAG,MAAM,CAAC,iBAAiB,CAAA;QACrD,CAAC;QACD,OAAO,EAAE,CAAC,EAAE,CAAC,EAAE,MAAM,EAAE,CAAA;IACxB,CAAC,CAAC,CAAA;IAEF,MAAM,QAAQ,GAAG,OAAO,CAAC,OAAO,EAAE,KAAK,EAAE,EAAE,SAAS,EAAE,OAAO,EAAE,MAAM,CAAC,OAAO,EAAE,CAAC,CAAA;IAEhF,uGAAuG;IACvG,qGAAqG;IACrG,wEAAwE;IACxE,MAAM,SAAS,GAAG,IAAI,GAAG,EAAwB,CAAA;IACjD,QAAQ,CAAC,OAAO,CAAC,CAAC,KAAK,EAAE,CAAC,EAAE,EAAE;QAC7B,KAAK,MAAM,MAAM,IAAI,KAAK;YAAE,SAAS,CAAC,GAAG,CAAC,MAAM,EAAE,CAAC,CAAC,CAAA;IACrD,CAAC,CAAC,CAAA;IACF,MAAM,cAAc,GAAG,IAAI,KAAK,CAAS,QAAQ,CAAC,MAAM,CAAC,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAA;IACxE,KAAK,MAAM,IAAI,IAAI,KAAK,EAAE,CAAC;QAC1B,IAAI,IAAI,CAAC,MAAM,GAAG,SAAS;YAAE,SAAQ;QACrC,MAAM,EAAE,GAAG,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAA;QAChC,IAAI,EAAE,KAAK,SAAS,IAAI,EAAE,KAAK,SAAS,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC;YAAE,SAAQ;QAC9D,IAAI,IAAI,CAAC,MAAM,GAAG,cAAc,CAAC,EAAE,CAAE;YAAE,cAAc,CAAC,EAAE,CAAC,GAAG,IAAI,CAAC,MAAM,CAAA;IACxE,CAAC;IAED,MAAM,QAAQ,GAAqB,QAAQ,CAAC,GAAG,CAAC,CAAC,KAAK,EAAE,CAAC,EAAE,EAAE;QAC5D,MAAM,GAAG,GAAG,cAAc,CAAC,KAAK,CAAC,IAAI,KAAK,CAAC,CAAC,CAAE,CAAA;QAC9C,OAAO;YACN,EAAE,EAAE,UAAU,CAAC,EAAE;YACjB,OAAO,EAAE,KAAK;YACd,cAAc,EAAE,GAAG;YACnB,UAAU,EAAE,GAAG,CAAC,OAAO,EAAE,OAAO,EAAE,UAAU,IAAI,SAAS;YACzD,QAAQ,EAAE,KAAK,CAAC,MAAM,GAAG,CAAC,IAAI,cAAc,CAAC,CAAC,CAAE,KAAK,QAAQ,CAAC,CAAC,CAAC,cAAc,CAAC,CAAC,CAAE,CAAC,CAAC,CAAC,IAAI;SACzF,CAAA;IACF,CAAC,CAAC,CAAA;IAEF,OAAO,EAAE,QAAQ,EAAE,cAAc,EAAE,KAAK,CAAC,MAAM,EAAE,aAAa,EAAE,CAAA;AACjE,CAAC"}

package/out/types.d.ts

@@ -21,6 +21,13 @@ export interface SourceRecord {
     address?: PostalAddress | null;
     phone?: string | null;
     email?: string | null;
+    /**
+     * Additional secondary-identifier fields, normalized — anything that helps tell two records apart
+     * or confirm they're the same beyond name/org/address/phone (an authorized-official name, a
+     * provider taxonomy, a license number…). Used as extra comparisons + corroborators when the model
+     * is built with matching `discriminators`. Keyed by a stable field name the model references.
+     */
+    attributes?: Record<string, string>;
     /** The original row, verbatim, for audit. */
     raw?: Record<string, string>;
 }

package/out/types.d.ts.map

@@ -1 +1 @@
-{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAEpF,oGAAoG;AACpG,MAAM,WAAW,YAAY;IAC5B,iEAAiE;IACjE,EAAE,EAAE,MAAM,CAAA;IACV,uDAAuD;IACvD,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACtB,wDAAwD;IACxD,IAAI,CAAC,EAAE,UAAU,GAAG,IAAI,CAAA;IACxB,kEAAkE;IAClE,YAAY,CAAC,EAAE,gBAAgB,GAAG,IAAI,CAAA;IACtC,oDAAoD;IACpD,OAAO,CAAC,EAAE,aAAa,GAAG,IAAI,CAAA;IAC9B,KAAK,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,KAAK,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,6CAA6C;IAC7C,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC5B;AAED,uGAAuG;AACvG,MAAM,WAAW,cAAc;IAC9B,yCAAyC;IACzC,EAAE,EAAE,MAAM,CAAA;IACV,uDAAuD;IACvD,OAAO,EAAE,YAAY,EAAE,CAAA;IACvB,qEAAqE;IACrE,cAAc,EAAE,YAAY,CAAA;IAC5B,gEAAgE;IAChE,UAAU,CAAC,EAAE;QAAE,QAAQ,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAA;KAAE,CAAA;IACpD;;;OAGG;IACH,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;CACvB;AAID,MAAM,WAAW,YAAY;IAC5B,IAAI,EAAE,OAAO,CAAA;IACb,qDAAqD;IACrD,WAAW,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC7B;AAED,MAAM,WAAW,cAAc;IAC9B,IAAI,EAAE,SAAS,CAAA;IACf,QAAQ,EAAE,YAAY,CAAA;IACtB,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;CACnC;AAED,MAAM,WAAW,wBAAwB;IACxC,IAAI,EAAE,mBAAmB,CAAA;IACzB,QAAQ,EAAE,cAAc,EAAE,CAAA;CAC1B"}
+{"version":3,"file":"types.d.ts","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,UAAU,EAAE,aAAa,EAAE,MAAM,mBAAmB,CAAA;AAEpF,oGAAoG;AACpG,MAAM,WAAW,YAAY;IAC5B,iEAAiE;IACjE,EAAE,EAAE,MAAM,CAAA;IACV,uDAAuD;IACvD,MAAM,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACtB,wDAAwD;IACxD,IAAI,CAAC,EAAE,UAAU,GAAG,IAAI,CAAA;IACxB,kEAAkE;IAClE,YAAY,CAAC,EAAE,gBAAgB,GAAG,IAAI,CAAA;IACtC,oDAAoD;IACpD,OAAO,CAAC,EAAE,aAAa,GAAG,IAAI,CAAA;IAC9B,KAAK,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB,KAAK,CAAC,EAAE,MAAM,GAAG,IAAI,CAAA;IACrB;;;;;OAKG;IACH,UAAU,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;IACnC,6CAA6C;IAC7C,GAAG,CAAC,EAAE,MAAM,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC5B;AAED,uGAAuG;AACvG,MAAM,WAAW,cAAc;IAC9B,yCAAyC;IACzC,EAAE,EAAE,MAAM,CAAA;IACV,uDAAuD;IACvD,OAAO,EAAE,YAAY,EAAE,CAAA;IACvB,qEAAqE;IACrE,cAAc,EAAE,YAAY,CAAA;IAC5B,gEAAgE;IAChE,UAAU,CAAC,EAAE;QAAE,QAAQ,EAAE,MAAM,CAAC;QAAC,SAAS,EAAE,MAAM,CAAA;KAAE,CAAA;IACpD;;;OAGG;IACH,QAAQ,EAAE,MAAM,GAAG,IAAI,CAAA;CACvB;AAID,MAAM,WAAW,YAAY;IAC5B,IAAI,EAAE,OAAO,CAAA;IACb,qDAAqD;IACrD,WAAW,EAAE,CAAC,MAAM,EAAE,MAAM,CAAC,CAAA;CAC7B;AAED,MAAM,WAAW,cAAc;IAC9B,IAAI,EAAE,SAAS,CAAA;IACf,QAAQ,EAAE,YAAY,CAAA;IACtB,UAAU,EAAE,MAAM,CAAC,MAAM,EAAE,OAAO,CAAC,CAAA;CACnC;AAED,MAAM,WAAW,wBAAwB;IACxC,IAAI,EAAE,mBAAmB,CAAA;IACzB,QAAQ,EAAE,cAAc,EAAE,CAAA;CAC1B"}

package/out/types.js.map

@@ -1 +1 @@
-{"version":3,"file":"types.js","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;;AA0DH,YAAY"}
+{"version":3,"file":"types.js","sourceRoot":"","sources":["../types.ts"],"names":[],"mappings":"AAAA;;;;;;;GAOG;;AAiEH,YAAY"}

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@mailwoman/registry",
-  "version": "4.8.1",
+  "version": "4.10.0",
   "description": "The geocode-first record-matching application: resolve messy contact/organization records into canonical, geocoded entities (block → score → cluster) and export them as GeoJSON for spatial analysis. The clinic-funding use case mailwoman was built for.",
   "license": "AGPL-3.0-only",
   "repository": {
@@ -17,9 +17,11 @@
     "./ingest": "./out/ingest.js"
   },
   "dependencies": {
-    "@mailwoman/match": "4.8.1",
-    "@mailwoman/record": "4.8.1",
-    "csv-parse": "^5.6.0"
+    "@mailwoman/address-id": "4.10.0",
+    "@mailwoman/match": "4.10.0",
+    "@mailwoman/record": "4.10.0",
+    "csv-parse": "^5.6.0",
+    "spliterator": "^2.0.0"
   },
   "devDependencies": {
     "@types/node": "^25.9.2"