SERVER-SIDE MATCH

• US 20080104043A1
• Filed: 10/25/2006
• Published: 05/01/2008
• 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
• Granted Patent: US 7,979,425 B2
• Time in Patent Office: Days
• US Class Current: 707/4
• 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 computer-implemented method for matching text queries to corresponding numerically queries comprising:

receiving a plurality of text queries at a server from a plurality of users;

generating a numerically equivalent query for each text query of the plurality of text queries by mapping symbols of the text query to associated numbers using a keypad standard that specifies associations between symbols and numbers;

generating a primary mapping between each numerically equivalent query and a text query having a substantially highest frequency of occurrence among text queries associated with the numerically equivalent query; and

outputting a primary mapped text query in response to receiving an associated numerically equivalent query transmitted from a user.

2. The method of claim 1, further comprising generating search results based on the primary mapped text query and transmitting the search results to the user that transmitted the numerically equivalent query.

3. The method of claim 1, further comprising generating additional mappings between the numerically equivalent query and additional text queries based on the whether a frequency of occurrence for the additional text queries exceeds a predetermined occurrence threshold.

4. The method of claim 3, further comprising outputting the additional text queries with the primary mapped text query.

5. The method of claim 1, wherein outputting comprises transmitting the primary mapped text query for display to the user.

6. The method of claim 1, further comprising generating a search result using the primary mapped text query, and transmitting the search result for display to the user.

7. The method of claim 3, wherein a first additional text query with a higher frequency of occurrence is mapped so that the first additional text query is highlighted in the output relative to a second additional text query with lesser frequency of occurrence.

8. The method of claim 1, wherein generating the primary mapping is performed on a recurrent basis so that the text query having the substantially highest frequency of occurrence can be updated if additional text queries are received at the server.

9. The method of claim 1, further comprising parsing a received text query comprising words into its constituent words and (treating) each constituent word as an individual text query.

10. The method of claim 1, wherein the numerically equivalent query transmitted by the user comprises a symbol indicating the numerically equivalent query includes more than one term.

11. The method of claim 10, further comprising determining that there is a previously generated numerically query that is substantially the same as the numerically equivalent query that includes the more than one term and selecting the corresponding primary mapped text query for outputting.

12. The method of claim 10, further comprising determining that there is not a previously generated numerically query that is substantially the same as the numerically equivalent query that includes the more than one term, selecting a primary mapped text query that corresponds to each term of the numeric equivalent query, and concatenating the primary mapped text queries for outputting.

13. The method of claim 12, wherein selecting the primary mapped text query that corresponds to each term of the numeric equivalents comprises favoring combinations of primary mapped text queries 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 used in generating numerically equivalent queries are associated with a particular search context.

16. The method of claim 15, wherein the search context is selected from a group consisting 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 associated with 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 generate numerically equivalent queries and to generate primary mappings.

19. The method of claim 1, wherein the user transmits the numerically equivalent query using a mobile device.

20. The method of claim 1, further comprising transmitting a query to correct a misspelling to the user if the numerically equivalent query transmitted from a user is associated with a primary mapped text query that comprises a misspelling.

21. The method of claim 1, further comprising using a stored profile, which includes text queries associated with the user, to weight frequency of occurrences for substantially similar text queries so that the substantially similar text queries are more frequently mapped to numerically equivalent query submitted by the user.

22. A system comprising:

means for generating a numerically equivalent query for each of a previously stored text query by mapping symbols of the text query to associated numbers using a keypad standard that specifies associations between numbers and symbols;

a frequency calculator to calculate a frequency of occurrence for each stored text query; and

a mapper to generate a mapping between the numerically equivalent query and a text query with a substantially highest frequency of occurrence among text queries associated with the numerically equivalent query.

23. A method for generating text queries based on numerically equivalent queries input by users, the method comprising:

generating numerically equivalent queries based on text queries by representing each symbol of a text query using a number;

mapping a text query having a greatest frequency of occurrence among the text queries corresponding to a numerically equivalent query; and

outputting the mapped text query in response to receiving the correspondent numerically equivalent query from a user.