Abstract: A spelling system derives a language model for a particular domain of structured data, the language model enabling determinations of alternative spellings of queries or other strings of text from that domain. More specifically, the spelling system calculates (a) probabilities that the various query entity types—such as STREET, CITY, or STATE for queries in the geographical domain—are arranged in each of the various possible orders, and (b) probabilities that an arbitrary query references given particular ones of the entities, such as the street “El Camino Real.” Based on the calculated probabilities, the spelling system generates a language model that has associated scores (e.g., probabilities) for each of a set of probable entity name orderings, where the total number of entity name orderings is substantially less than the number of all possible orderings.

Abstract: A system and computer-implemented method are provided for associating categories with business names for generalizing search queries, the method including identifying one or more businesses within a first geographic region, determining a business name and one or more categories for each of the one or more businesses, generating one or more name components for each of the one or more businesses from the name of the business, generating one or more name component groups from the name components of the one or more businesses, each name component group including one or more identical name components, determining for each name component group, if the one or more name components within the name component group are associated with businesses that share one or more common categories and associating the one or more common categories with the name component of the name component group.

Abstract: Provided is a process for identifying a new business listing, the process including: identifying, from a log of local search queries, a term that does not correspond to a name of a business listing; determining a number of recent search queries containing the term and a number of historical search queries containing the term; determining that a rate based on the number of recent search queries exceeds a threshold rate based on the number of historical search queries; and identifying the term as a name of a new business listing.

Abstract: A system and computer-implemented method are provided for associating categories with business names for generalizing search queries, the method including identifying one or more businesses within a first geographic region, determining a business name and one or more categories for each of the one or more businesses, generating one or more name components for each of the one or more businesses from the name of the business, generating one or more name component groups from the name components of the one or more businesses, each name component group including one or more identical name components, determining for each name component group, if the one or more name components within the name component group are associated with businesses that share one or more common categories and associating the one or more common categories with the name component of the name component group.

Abstract: Provided is a process of and apparatus for spatially indexing geographic items. The process may include obtaining geographic-item data identifying geographic items, the geographic location of each item, and an attribute of each item, wherein the geographic-item data identifies values of key-value pairs to be formed in a spatial index; obtaining a plurality of geographic-location keys each corresponding to a geographic area, the geographic-location keys identifying keys of the key-value pairs to be formed in the spatial index; and pairing each geographic-location key with an item among the geographic-item data. Pairing may be performed by: calculating distances between the geographic location of each of the items and the geographic-location key; weighting each of the distances based on the attribute of the item corresponding to that distance; and selecting the geographic item having the closest attribute-weighted distance as the item to be paired with the geographic-location key.

Abstract: A system and machine-implemented method for determining a language intent of a user submitted query is provided. A user query comprising text and a user location is received and a language usage signal based on the text of the user query is identified, wherein the language usage signal is associated with a first language. A second language associated with the user location is identified and a translation of the text of the user query from the first language to the second language is generated. A strength of the language usage signal is determined based on an amount of variation between the text of the user query and the translation of the text. When the strength of the language usage signal is greater than a predetermined threshold, an output language for the query is adjusted and results for the query are returned according to the output language.

Abstract: A spelling system derives a language model for a particular domain of structured data, the language model enabling determinations of alternative spellings of queries or other strings of text from that domain. More specifically, the spelling system calculates (a) probabilities that the various query entity types—such as STREET, CITY, or STATE for queries in the geographical domain—are arranged in each of the various possible orders, and (b) probabilities that an arbitrary query references given particular ones of the entities, such as the street “El Camino Real.” Based on the calculated probabilities, the spelling system generates a language model that has associated scores (e.g., probabilities) for each of a set of probable entity name orderings, where the total number of entity name orderings is substantially less than the number of all possible orderings.

Abstract: A system and machine-implemented method for determining a language intent of a user submitted query is provided. A user query comprising text and a user location is received and a language usage signal based on the text of the user query is identified, wherein the language usage signal is associated with a first language. A second language associated with the user location is identified and a translation of the text of the user query from the first language to the second language is generated. A strength of the language usage signal is determined based on an amount of variation between the text of the user query and the translation of the text. When the strength of the language usage signal is greater than a predetermined threshold, an output language for the query is adjusted and results for the query are returned according to the output language.