{"object":"match_recipes","pack_version":"2026.07.12","updated":"2026-07-12","entity_types":{"company":{"description":"Business / organization names (legal-form aware).","normalization":["unicode_fold","lower","strip_punct","legal_forms","business_stopwords"],"weights":{"token_set":0.45,"jaro_winkler":0.3,"phonetic":0.1,"edit":0.15},"default_threshold":82,"similarity_key":{"drop_stopwords":true,"sorted_tokens":true,"phonetic_first":true}},"person":{"description":"Individual person names (title/suffix/nickname aware).","normalization":["unicode_fold","lower","strip_punct","person_affixes","nicknames"],"weights":{"token_set":0.35,"jaro_winkler":0.3,"phonetic":0.2,"edit":0.15},"default_threshold":84,"similarity_key":{"drop_stopwords":false,"sorted_tokens":true,"phonetic_all":true}},"address":{"description":"Postal addresses (USPS-style abbreviation folding).","normalization":["unicode_fold","lower","strip_punct","address_abbrev"],"weights":{"token_set":0.55,"jaro_winkler":0.2,"phonetic":0.0,"edit":0.25},"default_threshold":80,"similarity_key":{"drop_stopwords":false,"sorted_tokens":true}},"product":{"description":"Product / SKU titles (light structural normalization).","normalization":["unicode_fold","lower","strip_punct"],"weights":{"token_set":0.5,"jaro_winkler":0.25,"phonetic":0.0,"edit":0.25},"default_threshold":85,"similarity_key":{"drop_stopwords":false,"sorted_tokens":true}},"org":{"description":"General organizations (alias of company pipeline).","normalization":["unicode_fold","lower","strip_punct","legal_forms","business_stopwords"],"weights":{"token_set":0.45,"jaro_winkler":0.3,"phonetic":0.1,"edit":0.15},"default_threshold":82,"similarity_key":{"drop_stopwords":true,"sorted_tokens":true,"phonetic_first":true}}},"default_type":"company","algorithms":{"token_set":"rapidfuzz.fuzz.token_set_ratio — order/duplicate-insensitive token overlap","jaro_winkler":"rapidfuzz.distance.JaroWinkler — prefix-weighted edit similarity","phonetic":"jellyfish/metaphone codes agree on the significant token (SOFT dep)","edit":"rapidfuzz.fuzz.ratio — Levenshtein-style character similarity"},"phonetic_backend":"jellyfish.metaphone","phonetic_available":true,"limits":{"max_value_len":400,"max_dedupe_items":1000},"licenses":[{"name":"rapidfuzz","kind":"library","license":"MIT","use":"token_set / jaro_winkler / edit similarity"},{"name":"jellyfish","kind":"library","license":"BSD/MIT","use":"Soundex / Metaphone phonetic codes (SOFT dependency)"},{"name":"metaphone","kind":"library","license":"BSD","use":"Double-Metaphone fallback (SOFT dependency)"},{"name":"unidecode","kind":"library","license":"GPL/Artistic","use":"unicode -> ASCII fold (SOFT dependency; stdlib NFKD fallback)"},{"name":"Company legal-form suffixes","kind":"reference-data","license":"public-domain (public facts)","use":"LEGAL_FORMS canonicalisation"},{"name":"USPS Publication 28 street-suffix abbreviations","kind":"reference-data","license":"public-domain (US government)","use":"ADDRESS_ABBREV folding"}],"disclaimer":"Structural normalization plus fuzzy matching over the caller's own strings: results are similarity indicators for deduplication and record-linkage, NOT authoritative identity resolution and NOT libpostal-grade international address parsing. Review before relying on a match.","note":"Structural normalization is applied BEFORE fuzzy matching (legal forms, business stop-words, address abbreviations, person affixes/nicknames). Freshness is a data update of the registry (no live daemon). NOT libpostal-grade international address parsing."}