Server-side match

• US 7,979,425 B2
• Filed: 10/25/2006
• Issued: 07/12/2011
• Est. Priority Date: 10/25/2006
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Datar, Mayur, Garg, Ashutosh
• Primary Examiner(s): Trujillo; James
• Assistant Examiner(s): Singh; Amresh
• Application Number:
  US11/552,751
• Publication Number:
  US 20080104043A1
• Time in Patent Office: 1,721 Days
• Field of Search: 707/100, 707/4, 707/721, 707/7
• US Class Current: 707/721
• CPC Class Codes: G06F 16/24534 Query rewriting; Transformation

Abstract

Systems and techniques for converting numeric queries into substantially equivalent textual queries are described. In general, the systems and techniques discussed use search query logs to accurately select a most probably mapping for a numeric-to-text conversion. This mapping can occur when a system (e.g., a server-side search system) receives a series of numeric inputs (e.g., from a cell phone keypad) that may correspond to more than one word. For example, a search server may receive input 22737, which corresponds to both the words ACRES and CASES, as part of a query. The server uses current entries in query logs to create mappings for words from the numeric input. If recent queries indicate that the term ACRES is currently more popular than the term CASES, the mapping may match the entry 22737 to the text ACRES.

Claims

1. A method, implemented by a server computer system, for matching text queries to corresponding numerical queries, the method comprising:

receiving, at the server computer system, a plurality of text queries, the text queries having been submitted to a search engine by a plurality of different users;

translating, by the server computer system, each text query of the plurality of text queries into a numerical representation, the translation including mapping text symbols that form each text query into numbers in accordance with a keypad standard that specifies associations between the text symbols and the numbers, wherein at least some of the text queries translate into the same numerical representation;

identifying, by the server computer system, a primary mapping for each numerical representation, the primary mapping being a single text query having a highest frequency of occurrence from amongst the text queries in the plurality of text queries that translate into the respective numerical representation; and

outputting, from the server computer system, the primary mapping of a particular numerical representation in response to receiving a numeric query that comprises the particular numerical representation from a remote device comprising a keypad mapped in accordance with the keypad standard.

2. The method of claim 1, further comprising:

generating, using a search engine, search results by searching with the primary mapping of the particular numerical representation; and

transmitting the search results to the remote device from which the numeric query was received.

3. The method of claim 1, further comprising identifying, by the server computer system, one or more additional mappings for the a first numerical representation, the additional mappings being additional text queries that translate into the first numerical representation, wherein the additional mappings are identified in response to determining that the frequencies at which the additional text queries occur in the plurality of text queries exceed a predetermined occurrence threshold.

4. The method of claim 3, further comprising outputting the additional mappings with the primary mapping.

5. The method of claim 3, further comprising displaying the additional mappings at the remote device, the displaying comprising highlighting a first additional mapping having a higher frequency of occurrence in the plurality of text queries that translate into the numerical representation in the output relative to a second additional mapping having a lesser frequency of occurrence in the plurality of text queries that translate into the numerical representation.

6. The method of claim 1, wherein outputting comprises transmitting the primary mapping to the remote device.

7. The method of claim 1, further comprising:

generating a search result by searching with the primary mapping to generate a search result; and

transmitting the search result to the remote device.

8. The method of claim 1, wherein the primary mappings are identified on a recurrent basis so that the text queries having the substantially highest frequency of occurrence in the plurality of text queries that translate into each respective numerical representation are updated.

9. The method of claim 1, further comprising:

parsing, by the server computer system, a text query in the plurality of text queries that comprises multiple words into its constituent words; and

treating each constituent word as an individual text query.

10. The method of claim 1, wherein the numeric query received from the remote device comprises a symbol indicating that the numeric query includes more than one term.

11. The method of claim 10, further comprising:

determining that a first text query has been previously translated into a first numerical representation that is substantially the same as the numeric query that includes the more than one term; and

in response thereto, selecting the primary mapping corresponding to the first numerical representation for the outputting.

12. The method of claim 10, further comprising:

determining that none of the plurality of text queries have been previously translated into a numerical representation that is substantially the same as the numeric query that includes more than one term;

selecting a respective primary mapping that corresponds to each term of the numeric equivalent query that includes more than one term; and

concatenating the respective primary mappings for the outputting.

13. The method of claim 12, wherein selecting the respective primary mapping that corresponds to each term of the numeric equivalents comprises favoring combinations of primary mappings that have a higher frequency of occurrence over combinations with a lesser frequency of occurrence.

14. The method of claim 1, wherein the keypad standard is selected from a group consisting of an International Keypad Standard, a North American Classic Keypad Standard, an Australian Classic Keypad Standard, a United Kingdom Classic Keypad Standard, and a Mobile 1 Keypad Standard.

15. The method of claim 1, wherein the received text queries are constrained to text queries submitted to the search engine in a particular search context.

16. The method of claim 15, wherein the search context is one of a news search context, a merchandise search context, a general web page search context, a location search context, an image search context, a news group search context, and a device search context.

17. The method of claim 1, wherein the received text queries used in generating numerically equivalent queries are constrained to text queries submitted to the search engine in a particular language or a particular country.

18. The method of claim 1, further comprising:

receiving text from a text corpus; and

using the received text to identify the numerical representations and to identify the primary mappings.

19. The method of claim 1, wherein the remote device is a mobile device.

20. The method of claim 1, further comprising transmitting, from the server computer system to the remote device, a query to correct a misspelling to the user if the primary mapping identified for the numeric query received from the remote device comprises a misspelling.

21. The method of claim 1, further comprising using a stored profile that includes text queries associated with a user of the remote device to weight frequencies of occurrences for text queries that are related to the text queries associated with the user so that the text queries that are related to the text queries associated with the user are output more frequently in response to numeric queries received from the user.

22. The method of claim 1, wherein:

the received plurality of text queries includes at least some multiple term search queries;

the multiple term search queries are each translated into a respective numerical representation that represents the entire multiple term search query, wherein at least some of the multiple term search queries translate into a same respective numerical representation.

23. The method of claim 22, wherein:

a first of two or more of the multiple term search queries that translate into the same respective numerical representation is identified as the primary mapping for the same respective numerical representation.

24. The method of claim 23, wherein:

the first of the multiple term search queries is output in response to the receipt of a first numeric query that comprises the same respective numerical representation.

25. A system comprising:

one or more computers having

means for generating a numerical representation for each of a collection of previously stored text queries by mapping symbols of the text queries into numbers in accordance with a keypad standard that specifies associations between numbers and symbols, the previously stored text queries having been submitted to a search engine by a plurality of different users, wherein at least some of the text queries generate the same numerical representation;

a frequency calculator programmed to calculate a frequency of occurrence for each stored text query in the collection of text queries; and

a mapper programmed to identify a primary mapping of a numeric query received from a remote device comprising a keypad mapped in accordance with the keypad standard, the primary mapping being a single text query of the collection having a highest frequency of occurrence in the collection from among the text queries of the collection for which the means for generating has generated a first numerical representation that matches the numeric query.

26. The system of claim 25, wherein the means for generating the numerical representations is programmed to:

generate numerical representations that each respectively represent an entire multiple term text query, wherein at least some of the multiple term text queries generate a same respective numerical representation.

27. The system of claim 26, wherein the means programmed to identify a first of the multiple term text queries that generate the same respective numerical representation as the primary mapping of a subsequent numeric query received from a remote device.

28. A method for matching text queries to corresponding numerical queries implemented by a server data processing system, the method comprising:

receiving, at the server system, a plurality of textual search queries submitted to a search engine by a plurality of different users;

translating, by the server system, each of the textual search queries into an ambiguous numerical representation using a keypad standard that specifies associations between symbols in the textual search queries and numbers, wherein at least some of the textual search queries translate into the same numerical representation;

receiving, at the server system, a current ambiguous numerical search query from a remote device comprising a keypad mapped to the keypad standard;

mapping, by the server system, the current ambiguous numerical search query to one or more current textual search queries using the translations of the textual search queries into the ambiguous numerical representations; and

outputting, from the server system, the one or more of the current textual search queries to which the current ambiguous numerical search query has been mapped.

29. The method of claim 28, wherein mapping the current ambiguous numerical search query to the one or more of the current textual search queries comprises:

identifying frequency rankings characterizing numbers of times each of the current textual search queries have been submitted to the search engine by the plurality of different users; and

mapping the current ambiguous numerical search query into the current textual search queries having a frequency ranking that exceeds a threshold and refraining to map the current ambiguous numerical search query into the current textual search queries having a frequency ranking that fails to exceed the threshold.

30. The method of claim 29, further comprising:

weighting the frequency rankings of the current textual search queries using a profile that profiles search queries associated with a user of the remote device, wherein the profile is stored in a data storage device accessible to the server system.

31. The method of claim 28, wherein outputting the one or more of the current textual search queries comprises transmitting a single primary mapped text query having the highest frequency ranking to the remote device and refraining to transmit other current textual search queries.

32. The method of claim 28, wherein:

the received plurality of textual search queries includes at least some multiple term textual search queries;

the multiple term textual search queries are each translated into a respective ambiguous numerical representation that represents the entirety of the multiple term textual search query, wherein at least some of the multiple term textual search queries translate into a same respective ambiguous numerical representation; and

a first of the multiple term textual search queries is output in response to the receipt of a current multiple term ambiguous numerical search query