摘要 |
Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating address component synonyms. In one aspect, a method includes determining that a plurality of addresses cannot be geocoded by a geocoding system. Variants of the addresses that can be geocoded by the geocoding system are generated, wherein each variant of a respective address lacks a removed term. Name terms for each variant are provided by the geocoding system. Each removed term is associated with name terms received for all variants that lack the removed term, including determining, for each associated name term of each removed term, a count of the number of variants for which the geocoding system provided the name term. Whether a name term is an address term synonym for a removed term is determined based at least in part on the count of the number of variants. |