Substitute term identification based on over-represented terms identification

• US 9,152,698 B1
• Filed: 01/03/2012
• Issued: 10/06/2015
• Est. Priority Date: 01/03/2012
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Dhamdhere, Kedar, Lamping, John Ogden, Tucker, Paul A.
• Primary Examiner(s): Hwa, Shyue Jiunn
• Application Number:
  US13/342,604
• Time in Patent Office: 1,372 Days
• Field of Search: 707/730, 704/1, 704/10, 704/260
• US Class Current: 1/1
• CPC Class Codes: G06F 16/332 Query formulation
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence

Abstract

Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for identifying substitute terms. According to one implementation, a method includes receiving an original query that includes one or more query terms; obtaining initial search results in response to the original query; identifying an over-represented term in text associated with a subset of the initial search results; determining that the over-represented term is associated with a particular query term; and in response to determining the over-represented term is associated with the particular query term, revising the original query to include the over-represented term.

Claims

1. A computer-implemented method comprising:

receiving an original query that includes one or more query terms;

determining, by one or more computers, not to apply a weak query term substitution rule to the original query, wherein the weak query term substitution rule identifies a particular term as a substitute for one or more of the query terms;

after determining not to apply the weak query term substitution rule to the original query, obtaining an initial set of search results from a text corpus of indexed resources;

determining, using the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight and by one or more computers, that the particular term occurs in text associated with a subset of the initial set of search results at a higher rate than the particular term occurs in the text corpus as a whole;

in response to determining that the particular term occurs in text associated with the subset of the initial set of search results at the higher rate than the particular term occurs in the text corpus as a whole, applying the weak query term substitution rule to the original query, to revise the original query to include the particular term; and

obtaining a subsequent set of search results in response to the revised query.

2. The method of claim 1, further comprising:

providing a subset of the subsequent set of search results.

3. The method of claim 1, wherein the initial set of search results are obtained from an initial text corpus of indexed resources and the subsequent set of search results are obtained from a different corpus of indexed resources.

4. The method of claim 3, wherein types of the different corpus of indexed resources comprise one or more of text, image, video, advertisement, and map.

5. The method of claim 1, wherein determining that the particular term occurs in text at the higher rate further comprises determining that the particular term occurs in text from one or more of titles, snippets, and resources associated with the subset of the initial set of search results.

6. The method of claim 1, wherein the weak query term substitution rule is associated with a low confidence score.

7. The method of claim 6, wherein the particular term is input of the weak query term substitution rule.

8. The method of claim 1, further comprising:

modifying a confidence score of the particular term associated with the one or more of the query terms.

9. The method of claim 1, wherein revising the original query to include the particular term further comprises adding the particular term and one or more logical operators to the original query.

10. The method of claim 1, wherein revising the original query to include the particular term further comprises replacing the one or more of the query terms with the particular term.

11. The method of claim 1, wherein determining that the particular term occurs in text further comprises determining that the particular term'"'"'s overall rate of occurrence in a document referenced by the initial set of search results is higher than the particular term'"'"'s rate of occurrence in the text corpus as a whole.

12. The method of claim 1, wherein determining that the particular term occurs in text associated with a subset of the initial set of search results at the higher rate further comprises determining that the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight is above a threshold.

13. A non-transitory computer-readable medium storing software having stored thereon instructions, which, when executed by one or more computers, cause the one or more computers to perform operations of:

receiving an original query that includes one or more query terms;

determining not to apply a weak query term substitution rule to the original query, wherein the weak query term substitution rule identifies a particular term as a substitute for one or more of the query terms;

after determining not to apply the weak query term substitution rule to the original query, obtaining an initial set of search results from a text corpus of indexed resources;

determining, using the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight, that the particular term occurs in text associated with a subset of the initial set of search results at a higher rate than the particular term occurs in the text corpus as a whole;

in response to determining that the particular term occurs in text associated with the subset of the initial set of search results at the higher rate than the particular term occurs in the text corpus as a whole, applying the weak query term substitution rule to the original query, to revise the original query to include the particular term; and

obtaining a subsequent set of search results in response to the revised query.

14. The computer-readable medium of claim 13, wherein the operations comprise:

providing a subset of the subsequent set of search results.

15. The computer-readable medium of claim 13, wherein the initial set of search results are obtained from an initial text corpus of indexed resources and the subsequent set of search results are obtained from a different corpus of indexed resources.

16. The computer-readable medium of claim 15, wherein types of the different corpus of indexed resources comprise one or more of text, image, video, advertisement, and map.

17. The computer-readable medium of claim 13, wherein determining that the particular term occurs in text at the higher rate further comprises determining that the particular term occurs in text from one or more of titles, snippets, and resources associated with the subset of the initial set of search results.

18. The computer-readable medium of claim 13, wherein the weak query term substitution rule is associated with a low confidence score.

19. The computer-readable medium of claim 18, wherein the particular term is input of the weak query term substitution rule.

20. The computer-readable medium of claim 13, wherein the operations comprise:

modifying a confidence score of the particular term associated with the one or more of the query terms.

21. The computer-readable medium of claim 13, wherein revising the original query to include the particular term further comprises adding the particular term and one or more logical operators to the original query.

22. The computer-readable medium of claim 13, wherein revising the original query to include the particular term further comprises replacing the one or more of the query terms with the particular term.

23. The computer-readable medium of claim 13, wherein determining that the particular term occurs in text further comprises determining that the particular term'"'"'s overall rate of occurrence in a document referenced by the initial set of search results is higher than the particular term'"'"'s rate of occurrence in the text corpus as a whole.

24. The computer-readable medium of claim 13, wherein determining that the particular term occurs in text associated with a subset of the initial set of search results at the higher rate further comprises determining that the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight is above a threshold.

25. A system comprising:

one or more computers and one or more storage devices storing instructions that are operable, when executed by the one or computers, to cause the one or more computers to perform operations comprising;

receiving an original query that includes one or more query terms;

determining not to apply a weak query term substitution rule to the original query, wherein the weak query term substitution rule identifies a particular term as a substitute for one or more of the query terms;

after determining not to apply the weak query term substitution rule to the original query, obtaining an initial set of search results from a text corpus of indexed resources;

determining, using the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight, that the particular term occurs in text associated with a subset of the initial set of search results at a higher rate than the particular term occurs in the text corpus as a whole;

in response to determining that the particular term occurs in text associated with the subset of the initial set of search results at the higher rate than the particular term occurs in the text corpus as a whole, applying the weak query term substitution rule to the original query, to revise the original query to include the particular term; and

obtaining a subsequent set of search results in response to the revised query.

26. The system of claim 25, wherein the operations comprise:

providing a subset of the subsequent set of search results.

27. The system of claim 25, wherein the initial set of search results are obtained from an initial text corpus of indexed resources and the subsequent set of search results are obtained from a different corpus of indexed resources.

28. The system of claim 27, wherein types of the different corpus of indexed resources comprise one or more of text, image, video, advertisement, and map.

29. The system of claim 25, wherein determining that the particular term occurs in text at the higher rate further comprises determining that the particular term occurs in text from one or more of titles, snippets, and resources associated with the subset of the initial set of search results.

30. The system of claim 25, wherein the weak query term substitution rule is associated with a low confidence score.

31. The system of claim 30, wherein the particular term is input of the weak query term substitution rule.

32. The system of claim 25, wherein the operations further comprise:

modifying a confidence score of the particular term associated with the one or more of the query terms.

33. The system of claim 25, wherein revising the original query to include the particular term further comprises adding the particular term and one or more logical operators to the original query.

34. The system of claim 25, wherein revising the original query to include the particular term further comprises replacing the one or more of the query terms with the particular term.

35. The system of claim 25, wherein determining that the particular term occurs in text further comprises determining that the particular term'"'"'s overall rate of occurrence in a document referenced by the initial set of search results is higher than the particular term'"'"'s rate of occurrence in the text corpus as a whole.

36. The system of claim 25, wherein determining that the particular term occurs in text associated with a subset of the initial set of search results at the higher rate further comprises determining that the particular term'"'"'s frequency-inverse document frequency (tf-idf) weight is above a threshold.