Skip to main content

Falsehoods programmers believe about street names

Falsehoods​

Programmers consistently assume streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. have type suffixes, don't contain numbers, are unique within a citylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy., and that every address has one. Real streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. violate all of these.

"Street names end in descriptors like 'street', 'avenue', or 'drive'."​

Piccadilly, London, W1J 9PN has no type suffix. Neither does Broadway, New York, NY nor The Strand, London. Many of the world's most famous streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. lack the suffix pattern that address validators demand.

PeliasPeliasAn open-source geocoder, Mailwoman's spiritual predecessor.'s street classifier uses a suffix dictionary to identify streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. tokenstokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words.: if a tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. ends in "StreetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.," "Avenue," "Road," "Boulevard," etc., it's likely part of a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name. Without a suffix, the classifier relies on position (tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. after a house numberhouse numberThe numeric or alphanumeric identifier of a building on a street. Mailwoman's house_number component; its position relative to the street name flips between locales.) and neighboring components. Piccadilly alone, without a postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. or localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. context, could be a venuevenueA named, non-address place — a business, building, park, or stadium. Mailwoman's free-text point-of-interest component, added as a Tier 2 fine label., a localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy., or a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. — PeliasPeliasAn open-source geocoder, Mailwoman's spiritual predecessor. has no way to distinguish.

"When they do have a descriptor, there's only one."​

17 Hill Street, London, W1J 5LJ has "Hill" (which can be a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. type) and "StreetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.." Avenue Road, Toronto, Ontario is an entire streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name composed of two streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. type words. A parser that strips suffixes will reduce Avenue Road to nothing.

"The descriptor is at the end."​

French addresses place the descriptor at the beginning: rue de Rivoli, avenue des Champs-Élysées, place de la Concorde. Spanish: Calle Mayor. Italian: Via Roma. German: Hauptstraße. The Anglophone "StreetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. at the end" pattern is the exception, not the rule.

This is a structural problem for rule-based classifiersrule-based classifierMailwoman's legacy v0 parser — a library of deterministic token classifiers (house number, street suffix, postcode, place name, etc.) composed by priority. Now primarily used for corpus labeling, fallback classification, and arbitration diagnostics.. A US-trained parser expects [number] [street name] [street type] [city] [state] [zip]. A French address is [number] [street type] [street name] [postcode] [city]. Every positional assumption is wrong. The parser must either maintain per-localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for. ordering rules or learn ordering from data.

"A street name won't include a number."​

8 Seven Gardens Burgh, WOODBRIDGE, IP13 6SU. Plein 1944, Nijmegen, Netherlands — streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. can be named after years, and those years look like building numbers. When the parser sees Plein 1944 85 (streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. Plein 1944, building 85), it must distinguish the year-in-streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.-name from the building number — two adjacent numbers with different semantic roles.

"When a numbered street and a building number are adjacent, there's a separator."​

Dutch: Gondel 2695, Lelystad — area Gondel, streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. 26, number 95. No separator. The string "2695" is both a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. identifier and a building identifier packed into one tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words.. A rule-based parser sees one number and classifies it as one thing — but it's two.

"Street names don't recur in the same city."​

London has seventeen different High Streets. Without a postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one., 10 High Street, London is ambiguous across seventeen locations. The resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon. needs the postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. or localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. to disambiguate, but the parser may have already assigned locality=London to all seventeen.

This is the same problem as duplicate localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. names (Springfield, Paris) but at a finer grain. The resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon.'s concordanceconcordanceA joint-decode signal that rewards a parse whose spans form a consistent Who's On First parent-child chain and vetoes contradictory ones — a hard veto for conflicts, a log-space bonus for full agreement. scoring (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 5) handles this by checking whether a (street, locality, region, postcode) tuple is jointly coherent in WOFWOF (Who's On First). An open-source gazetteer of places maintained by Mapzen/whosonfirst. Mailwoman builds a custom SQLite database from WOF GeoJSON repos, extended with postcode data, importance scores, and coincident-role relations. — but WOF does not index street names, only administrative places. StreetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. duplication is outside the resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon.'s concordanceconcordanceA joint-decode signal that rewards a parse whose spans form a consistent Who's On First parent-child chain and vetoes contradictory ones — a hard veto for conflicts, a log-space bonus for full agreement. reach.

"A road has exactly one name."​

The A1 in the UK is a 410-mile road composed of Goswell Road, Regent Road, and dozens of other named segmentssegmentA punctuation-bounded chunk of the normalized input — the comma-separated parts of 'Portland, OR' — used to give downstream stages structural context.. Multiple buildings numbered 1 exist on different segmentssegmentA punctuation-bounded chunk of the normalized input — the comma-separated parts of 'Portland, OR' — used to give downstream stages structural context.. The road's official designation (A1) is not the streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name anyone uses.

"Addresses have exactly one street."​

Royal Mail "dependent streetsdependent streetA secondary street name required for delivery in some postal systems (notably Royal Mail), as in '6 Elm Avenue, Runcorn Road, Birmingham' where Runcorn Road is the dependent street.": 6 Elm Avenue, Runcorn Road, Birmingham, B12 8QX. Runcorn Road is the main streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.; Elm Avenue is a stub that isn't unique within the citylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy.. The address requires both for delivery.

"Addresses have a street at all."​

Japan does not use named streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. for most addresses: 東京都千代田区丸の内1-1-1 is district Marunouchi, block 1, building 1 — three levels of block-based numbering, no streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name. Rural US routes: Box 1234, R.R. 1, Winthrop, ME 04364 — a box on a route, no streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.. Nicaragua: From where the Chinese restaurant used to be, two blocks down — landmarks and directions, no streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels..

Mannheim, Germany uses a grid system: R 5, 6-13, D-68161 Mannheim — block R, row 5, buildings 6-13, with no streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name or streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. type at all.

How traditional geocoders handled these​

libpostallibpostalAn open-source C address parser used by Pelias. Mailwoman's rule-based v0 and neural classifier supersede it. handles suffix-less streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. through its CRFCRF (Conditional Random Field). A statistical modeling method that predicts structured outputs by modeling dependencies between adjacent labels. Mailwoman uses a linear-chain CRF as the Viterbi decoder at inference time to enforce BIO label consistency — a B-street must be followed by I-street or O, never I-locality.'s transition modelneural classifierThe machine learning model at the core of Mailwoman's parser — a transformer encoder (~30M parameters) trained from scratch to do BIO token classification over addresses. It learns the 'grammar' of address formats; the gazetteer supplies the 'atlas.': a tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. between a house numberhouse numberThe numeric or alphanumeric identifier of a building on a street. Mailwoman's house_number component; its position relative to the street name flips between locales. and a localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. is likely a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels., even without a suffix. It handles prefix-descriptor streetsstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. (European) through localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for.-specific trainingtrainingThe process of adjusting a model's parameters so its predictions match labeled examples, by repeatedly measuring error and nudging the weights to reduce it. Distinct from inference, when the trained model is run on new input. — the French modelneural classifierThe machine learning model at the core of Mailwoman's parser — a transformer encoder (~30M parameters) trained from scratch to do BIO token classification over addresses. It learns the 'grammar' of address formats; the gazetteer supplies the 'atlas.' learns rue as a street prefixstreet affixA modifier on a street name indicating type or direction — Street, Avenue, rue, Calle, N, East. Mailwoman tags these as street_prefix / street_suffix, recognized via a morphology FST., the US modelneural classifierThe machine learning model at the core of Mailwoman's parser — a transformer encoder (~30M parameters) trained from scratch to do BIO token classification over addresses. It learns the 'grammar' of address formats; the gazetteer supplies the 'atlas.' learns Street as a suffix. But libpostallibpostalAn open-source C address parser used by Pelias. Mailwoman's rule-based v0 and neural classifier supersede it.'s per-localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for. modelsneural classifierThe machine learning model at the core of Mailwoman's parser — a transformer encoder (~30M parameters) trained from scratch to do BIO token classification over addresses. It learns the 'grammar' of address formats; the gazetteer supplies the 'atlas.' are separate binaries; a single deployment serving multiple localeslocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for. needs to know which modelneural classifierThe machine learning model at the core of Mailwoman's parser — a transformer encoder (~30M parameters) trained from scratch to do BIO token classification over addresses. It learns the 'grammar' of address formats; the gazetteer supplies the 'atlas.' to invoke.

PeliasPeliasAn open-source geocoder, Mailwoman's spiritual predecessor. uses a suffix dictionary with localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for.-specific entries. street classification in PeliasPeliasAn open-source geocoder, Mailwoman's spiritual predecessor. relies on suffix matching plus adjacency to known components. Piccadilly without a suffix would be classified as street only if it appears between a house numberhouse numberThe numeric or alphanumeric identifier of a building on a street. Mailwoman's house_number component; its position relative to the street name flips between locales. and a localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy. — a positional heuristic that fails when the address has non-standard ordering.

Google's API handles most of these cases well in practice. It correctly parsesaddress parsingThe process of decomposing a free-text postal address string into structured components — house number, street name, locality, region, postcode, and country — so a geocoder can resolve them to coordinates. rue de Rivoli and Japanese block addresses. But the parser is proprietary — you cannot inspect how it handles the Dutch Gondel 2695 case or whether it distinguishes Ten Post Office Sq from 10 Post Office Sq.

What the neural approach changes​

The staged pipelinestaged pipelineMailwoman's runtime architecture: a sequence of pure-function stages (normalize → query-shape → locale-gate → kind-classifier → phrase-grouper → classifier → decoder) connected by typed handoffs. Each stage is published as its own npm package. addresses the falsehoods in layerslayerOne transformer block — attention plus a feed-forward network, with normalization and residual connections — applied to every position. Stacking layers lets the model build up richer representations; Mailwoman's encoder has 6.:

Locale gatelocale gateStage 2 of the runtime pipeline: rule-based locale detection from the query shape's script and known-format signals. Returns a LocaleHint with the top candidate and alternatives, surfacing disagreement with an explicit --locale flag. (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 2) detects the language/script family before classification. A French address triggers French ordering expectations (postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. before localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy., streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. type before streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. name). A Japanese address skips streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. classification entirely and routes to block-based component tagscomponent tagOne of the 33 labels in Mailwoman's address schema — street, locality, region, postcode, house_number, unit, po_box, country, venue, intersection, and others. Each parsed span carries exactly one component tag.. The classifier does not need to handle every localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for.'s ordering — it only needs to handle the localelocaleThe combination of language and country an address comes from. en-US and fr-FR are the locales Mailwoman ships weights for. the gate has detected.

Phrase grouperphrase grouperStage 2.7 of the runtime pipeline: proposes coherent input units (street phrase, locality phrase, postcode, etc.) with structural kind hypotheses. Decouples boundary discovery from type classification so the classifier answers 'what type?' not 'where?' (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 2.7) proposes spansspanA contiguous range of characters or tokens in the input string, tagged with an address component type (street, locality, postcode, etc.). Parsed addresses are represented as collections of spans, possibly nested in a tree. based on structural cues, not dictionaries. Gondel 2695 would be proposed as a single spanspanA contiguous range of characters or tokens in the input string, tagged with an address component type (street, locality, postcode, etc.). Parsed addresses are represented as collections of spans, possibly nested in a tree. (no whitespace, no punctuation boundary) — the grouper does not need to know that "26" and "95" have different semantic roles. The classifier then types the spanspanA contiguous range of characters or tokens in the input string, tagged with an address component type (street, locality, postcode, etc.). Parsed addresses are represented as collections of spans, possibly nested in a tree., and the Joint Decoder (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 5) can split it if needed.

The classifier (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 3) learns streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.-name distributions from corpuscorpusThe BIO-labeled training data used to train Mailwoman's neural classifier. Assembled from real sources (OpenAddresses, National Address Database) and synthetic shards (boundary stress, order variants, negative space). Managed by @mailwoman/corpus. co-occurrence. A tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. following a house-number-shaped tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. and preceding a localitylocalityThe city / town / settlement component of an address: a populated place sitting between region and neighbourhood in the hierarchy.-shaped tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. is likely a streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels., regardless of suffix. A tokentokenOne word or subword in the tokenized input. For the neural classifier, tokens come from SentencePiece (subword units); for the rule classifiers, tokens are whitespace- and punctuation-separated words. like Piccadilly that appears frequently in streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. contexts will be classified as street even without a suffix — not because it matches a dictionary, but because the trainingtrainingThe process of adjusting a model's parameters so its predictions match labeled examples, by repeatedly measuring error and nudging the weights to reduce it. Distinct from inference, when the trained model is run on new input. data consistently labelscomponent tagOne of the 33 labels in Mailwoman's address schema — street, locality, region, postcode, house_number, unit, po_box, country, venue, intersection, and others. Each parsed span carries exactly one component tag. it as street.

The reconciler (StagestageOne of the dataflow stages in the runtime pipeline (normalize, locale gate, kind classify, phrase group, token classify, sequence correct, reconcile, resolve). Distinct from tier (model vocabulary) and phase (plan milestone). 5) catches the "seventeen High Streets in London" problem by checking resolved candidates: if 10 High Street resolves to seventeen different WOFWOF (Who's On First). An open-source gazetteer of places maintained by Mapzen/whosonfirst. Mailwoman builds a custom SQLite database from WOF GeoJSON repos, extended with postcode data, importance scores, and coincident-role relations. administrative areas within London, the reconciler cannot pick one without additional context (postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one., boroughboroughAn administrative or historical division of a city — e.g. the five boroughs of New York City. May be postal, legal, or both, and complicates the locality hierarchy.). It surfaces the ambiguity rather than picking one.

What Mailwoman still can't do​

  • Dependent streetsdependent streetA secondary street name required for delivery in some postal systems (notably Royal Mail), as in '6 Elm Avenue, Runcorn Road, Birmingham' where Runcorn Road is the dependent street.. The schema has no dependent_street tag. 6 Elm Avenue, Runcorn Road would parseaddress parsingThe process of decomposing a free-text postal address string into structured components — house number, street name, locality, region, postcode, and country — so a geocoder can resolve them to coordinates. as street=Runcorn Road with Elm Avenue either dropped or classified as a second streetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels.. The resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon. cannot use dependent-street relationships because WOFWOF (Who's On First). An open-source gazetteer of places maintained by Mapzen/whosonfirst. Mailwoman builds a custom SQLite database from WOF GeoJSON repos, extended with postcode data, importance scores, and coincident-role relations. doesn't encode them.
  • StreetstreetThe named linear feature along which house numbers are ordered. Decomposes into a name plus street affixes; one of the Tier 2 fine labels. disambiguation without postcodespostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one.. 10 High Street, London is structurally ambiguous and the resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon. cannot resolve it without a postcodepostcodeThe country-specific postal code (US ZIP, French code postal, etc.). Mailwoman handles postcode parsing entirely by rule classifier — a regex problem, not an ML one. or boroughboroughAn administrative or historical division of a city — e.g. the five boroughs of New York City. May be postal, legal, or both, and complicates the locality hierarchy.. The parser correctly identifies the components; the resolverresolverThe component that converts parsed address components (locality, region, postcode) into coordinates by looking them up in the gazetteer. The resolver ranks candidates by name match, population, and proximity, and returns the best-matching place with its centroid or polygon. correctly identifies the ambiguity. The downstream application must decide.
  • Grid-based addressing. Mannheim's R 5, 6-13 requires block-and-row tags that Mailwoman's current schema does not have. Japan's 1-1-1 format is supported through the JP-specific tags (district, block, sub_block, building_number) but these are deferred to PhasephaseA milestone in the implementation plan (Foundation, Corpus, Training, Integration, and forward-looking phases). Distinct from stage (runtime pipeline) and tier (model vocabulary). 6.

References​