INFORMATION RETRIEVAL USING CATEGORY AS A CONSIDERATION

• US 20090125505A1
• Filed: 04/25/2008
• Published: 05/14/2009
• Est. Priority Date: 11/13/2007
• Status: Active Grant

• Current Assignee: Walmart Apollo, LLC (WalMart Inc.)
• Original Assignee: Kosmix Corporation
• Inventors: Bhalotia, Gaurav, Adams, John Patrick
• Granted Patent: US 8,862,608 B2
• Time in Patent Office: Days
• US Class Current: 1/1
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 16/367 Ontology
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

Category affinity may be used as a consideration in providing search results. A taxonomy of substantive categories is created and/or obtained. A corpus of document is compared with the taxonomy to determine the category(ies) with which the documents affine. A query is also compared with the taxonomy to determine the category(ies) with which the query affines. A document may receive a category score based on how well the document'"'"'s category(ies) match the query'"'"'s category(ies). This document score may be combined with other scores, such as a text score, a link score, and a distance score, and/or any other factors, to determine an overall relevance score. The relevance score may then be used to rank and present search results.

Claims

1. A method of providing search results based on a query, the method comprising:

creating a content/category affinity by determining which one or more categories each of a plurality of documents in a corpus relates to;

creating a query/category affinity by determining which one or more categories the query relates to;

making an assessment of relevance of said documents to said query based on one or more factors that comprise;

which of said documents have at least one category in common with said query; and

providing results to a person based on said assessment.

2. The method of claim 1, wherein each of the one or more categories is associated with one or more features relating to the category, wherein said making of said assessment comprises:

tagging said document with a particular category based on a finding that the document contains more features associated with the particular category than with any other category.

3. The method of claim 1, wherein the one or more factors on which said assessment is based further comprise:

a comparison of terms in the documents with one or more terms in the query.

4. The method of claim 1, wherein data associates at least some of the documents in the corpus with one or more geographic locations, wherein the query comprises an indication of a first geographic location, and wherein the one or more factors on which said assessment is based further comprise:

a comparison of said one or more geographic locations with said first geographic location.

5. The method of claim 1, wherein the one or more factors on which said assessment is based further comprise:

a number of referring documents that link to each of said documents; and

an assessment of how relevant each of the referring documents is to a document that the referring document links to.

6. The method of claim 1, further comprising:

creating a taxonomy that comprises said categories, which associates each of said categories with one or more features.

7. The method of claim 6, wherein said creating said content/category affinity comprises determining, for each document in the corpus, which one or more categories are associated with features that appear in the document.

8. The method of claim 1, further comprising:

creating at least one dictionary that comprises a plurality of words and that associates each word with one or more of said categories,

wherein determining which one or more categories a document in the corpus relates to, or which one or more categories the query relates to, is performed by comparing words in the document or query with words in the at least one dictionary.

9. The method of claim 8, wherein said categories are organized into a multi-level hierarchy in which more general categories are at a relatively higher level and more specific categories included in the more general categories are at a relatively lower level, and wherein said at least one dictionary comprises:

a first dictionary that associates words with categories at a first level of granularity corresponding to a first level in the hierarchy; and

a second dictionary that associates words with categories at a second level of granularity corresponding to a second level in the hierarchy.

10. The method of claim 8, wherein said categories are organized into a multi-level hierarchy that places different businesses into said categories, and wherein a category is placed at a particular level in a hierarchy based on the category having approximately the same number of businesses as other categories in that particular level.

11. The method of claim 1, wherein said making of said assessment comprises:

determining a statistical surprisingness of said query'"'"'s association with a particular category.

12. The method of claim 11, wherein said determining of said statistical surprisingness comprises:

determining a first number of documents that match said query based on terms in the query and the documents;

determining a second number of documents that match each of the one or more categories; and

performing an evaluation of the query'"'"'s likelihood of matching the category based on said first number and said second number.

13. The method of claim 1, wherein said providing of said results to said person comprises:

transmitting search results to said person'"'"'s computing system to be displayed on a display connected to said computing system.

14. One or more computer-readable storage media comprising executable instructions to perform a method of providing search results based on a query, the method comprising:

creating a content/category affinity by determining which one or more categories each of a plurality of documents in a corpus relates to;

creating a query/category affinity by determining which one or more categories the query relates to;

making an assessment of relevance of said documents to said query based on one or more factors that comprise;

which of said documents have at least one category in common with said query; and

providing results to a person based on said assessment.

15. The one or more computer-readable storage media of claim 14, wherein each of the one or more categories is associated with one or more features relating to the category, wherein said making of said assessment comprises:

tagging said document with a particular category based on a finding that the document contains more features associated with the particular category than with any other category.

16. The one or more computer-readable storage media of claim 14, wherein data associates at least some of the documents in the corpus with one or more geographic locations, wherein the query comprises an indication of a first geographic location, and wherein the one or more factors on which said assessment is based further comprise:

a comparison of said one or more geographic locations with said first geographic location.

17. The one or more computer-readable storage media of claim 14, wherein the method further comprises:

creating a taxonomy that comprises said categories, which associates each of said categories with one or more words.

18. The one or more computer-readable storage media of claim 14, wherein the method further comprises:

creating at least one dictionary that comprises a plurality of words and that associates each word with one or more of said categories,

wherein determining which one or more categories a document in the corpus relates to, or which one or more categories the query relates to, is performed by comparing words in the document or query with words in the at least one dictionary.

19. The one or more computer-readable storage media of claim 14, wherein said making of said assessment comprises:

determining a statistical surprisingness of said query'"'"'s association with a particular category.

20. A system for providing search results based on a query, the system comprising:

means for creating a content/category affinity by determining which one or more categories each of a plurality of documents in a corpus relates to;

means for creating a query/category affinity by determining which one or more categories the query relates to;

means for making an assessment of relevance of said documents to said query based on one or more factors that comprise;

which of said documents have at least one category in common with said query; and

means for providing results to a person based on said assessment.