Classification of ambiguous geographic references

• US 20060149742A1
• Filed: 12/30/2004
• Published: 07/06/2006
• Est. Priority Date: 12/30/2004
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Egnor, Daniel
• Granted Patent: US 7,716,162 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/285 Clustering or classification
• CPC Class Codes: G06F 16/29 Geographical information databases
• CPC Class Codes: G06F 16/31 Indexing; Data structures therefor; Storage structures
• CPC Class Codes: G06F 16/9537 Spatial or temporal dependent retrieval, e.g. spatiotemporal queries
• CPC Class Codes: H04L 67/10 in which an application is distributed across nodes in the network software deployment G06F8/60; multiprogramming arrangements G06F9/46
• CPC Class Codes: Y10S 707/99935 Query augmenting and refining, e.g. inexact access

Abstract

A location classifier generates location information based on textual strings in input text. The location information defines potential geographical relevance of the input text. In determining the location information, the location classifier may receive at least one geo-relevance profile associated with at least one string in the input text, obtain a combined geo-relevance profile for the document from the at least one geo-relevance profile, and determine geographical relevance of the input text based on the combined geo-relevance profile.

Claims

1. A method of determining geographical relevance of a document comprising:

receiving at least one geo-relevance profile associated with at least one string in the document;

obtaining a combined geo-relevance profile for the document from the at least one geo-relevance profile; and

determining geographical relevance of the document based on the combined geo-relevance profile.

2. The method of claim 1, wherein the at least one geo-relevance profile includes a plurality of geo-relevance profiles and each of the plurality of geo-relevance profiles are histograms relating occurrences of one of the strings to geographical regions.

3. The method of claim 2, wherein obtaining the combined geo-relevance profile includes:

multiplying the plurality of histograms.

4. The method of claim 1, wherein determining geographical relevance of the document based on the combined geo-relevance profile includes:

analyzing the combined geo-relevance profile for peaks.

5. The method of claim 1, wherein receiving at least one geo-relevance profile includes:

examining the document to locate the at least one string by comparing strings in the document to a plurality of strings for which geo-relevance profiles were previously generated.

6. The method of claim 1, wherein the at least one string is a single term string.

7. The method of claim 1, wherein the at least one string includes bi-grams.

8. The method of claim 1, wherein the document is a web page.

9. The method of claim 1, wherein the document is a search query.

10. A location classifier comprising:

means for receiving input text;

means for locating strings within the input text that were previously determined to be geographically relevant;

means for retrieving geo-relevance profiles for the located strings; and

means for determining whether the input text is geographically relevant based on the retrieved geo-relevance profiles.

11. The location classifier of claim 10, further comprising:

means for combining the retrieved geo-relevance profiles.

12. A computer-readable medium containing programming instructions for execution by a processor, the computer-readable medium comprising:

programming instructions for receiving a plurality of geo-relevance profiles associated with a respective plurality of strings in a document, the geo-relevance profiles each defining the geographical relevance of the string with respect to geographical regions; and

programming instructions for determining geographical relevance of the document based on the plurality of geo-relevance profiles.

13. The computer-readable medium of claim 12, further including:

programming instructions for combining the plurality of geo-relevance profiles to obtain a combined geo-relevance profile for the document.

14. The computer-readable medium of claim 13, wherein the programming instructions for combining the geo-relevance profiles include:

programming instructions for multiplying the plurality of geo-relevance profiles together.

15. The computer-readable medium of claim 13, wherein the programming instructions for determining geographical relevance of the document further includes:

programming instructions for analyzing the combined geo-relevance profile for peaks.

16. The computer-readable medium of claim 12, wherein the plurality of geo-relevance profiles include histograms.

17. The computer-readable medium of claim 12, wherein the strings include single term strings.

18. The computer-readable medium of claim 12, wherein the strings include bi-grams.

19. The computer-readable medium of claim 12, wherein the document is a web page.

20. The computer-readable medium of claim 12, wherein the document is a search query.

21. A method for generating a geo-relevance profile for a string, the method comprising:

determining a plurality of sections of training text in which each section of training text is associated with a geographical region;

accumulating occurrences of the string in the plurality of selections of training text; and

generating the geo-relevance profile as a histogram based on the accumulated occurrences of the string.

22. The method of claim 21, wherein the histogram relates the occurrences of the string to geographical regions.

23. The method of claim 21, wherein determining a plurality of sections of training text includes:

locating known geographic signals in a plurality of documents; and

selecting the sections of training text from those of the plurality of documents that include the known geographic signals.

24. The method of claim 21, wherein the documents are web documents.

25. The method of claim 21, further comprising:

normalizing the generated geo-relevance profile based on a global distribution of geographical regions within the training text.

26. The method of claim 21, wherein the geographical regions are represented by zip codes.

27. The method of claim 21, further comprising:

saving the generated geo-relevance profile for future use when the geo-relevance profile includes statistically relevant peaks.

28. The method of claim 21, wherein accumulating occurrences of the string further includes:

not accumulating occurrences of the string when textual context surrounding the string matches the textual context surrounding a previous occurrence of the string.

29. A device comprising:

a processor; and

a computer-readable memory coupled to the processor and containing instructions that when executed by the processor cause the processor to;

determine a plurality of sections of training text in which each section of training text is associated with a geographical region;

accumulate occurrences of a string in the plurality of selections of training text; and

generate a geo-relevance profile that defines the geographical relevance of the string with respect to geographical regions based on the accumulated occurrences of the string.

30. A method comprising:

examining a document to locate strings that were previously determined to have geographical relevance;

receiving geo-relevance strings associated with the strings;

obtaining a combined geo-relevance profile for the document from the received geo-relevance profiles by multiplying the received geo-relevance profiles together;

analyzing the combined geo-relevance profile for peaks; and

determining geographical relevance of the document based on whether peaks are present in the combined geo-relevance profile.