Item recommendations using keyword expansion

• US 8,615,524 B2
• Filed: 01/26/2012
• Issued: 12/24/2013
• Est. Priority Date: 05/25/2007
• Status: Active Grant

• Current Assignee: PRJ Holding Company LLC
• Original Assignee: Piksel Incorporated
• Inventors: Kanigsberg, Issar Amit, Veidlinger, Daniel Marc, El Shazli, Tamer, Mozersky, Myer Joshua
• Primary Examiner(s): Fleurantin, Jean B
• Application Number:
  US13/359,288
• Publication Number:
  US 20120330992A1
• Time in Patent Office: 698 Days
• Field of Search: 707/706, 707/707, 707/765, 707/769, 707/770, 707/776, 707/796
• US Class Current: 707/765
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising
• CPC Class Codes: G06Q 30/0255 based on user history
• CPC Class Codes: G06Q 30/0269 based on user profile or attribute
• CPC Class Codes: G06Q 30/0625 Directed, with specific intent or strategy
• CPC Class Codes: G06Q 30/0631 Item recommendations

Abstract

A search technology generates recommendations with minimal user data and participation, and provides better interpretation of user data, such as popularity, thus obtaining breadth and quality in recommendations. It is sensitive to the semantic content of natural language terms and lets users briefly describe the intended recipient (i.e., interests, eccentricities, previously successful gifts). Based on that input, the recommendation software system and method determines the meaning of the entered terms and creatively discover connections to gift recommendations from the vast array of possibilities. The user may then make a selection from these recommendations. The search/recommendation engine allows the user to find gifts through connections that are not limited to previously available information on the Internet. Thus, interests can be connected to buying behavior by relating terms to respective items.

Claims

1. A computer implemented method of recommending products and services executed by one or more computer processors, comprising the digital processing steps of:

providing an item recommendation for a subject user profile, where each item has a respective set of associated keywords, the item recommendation being provided by;

processing the subject user profile to extract an initial set of keywords;

comparing the initial set of extracted user profile keywords with co-occurring keywords identified in a stored data expansion model that is configured to provide the measured degree to which at least two keywords tend to occur together in a multitude of user profiles;

expanding the initial set of extracted user profile keywords with additional related keywords, where the additional related keywords are determined using one or more of the identified co-occurring keywords from the expansion model that frequently appear together with the extracted user profile keywords in at least a portion of the multitude of user profiles, the expansion model being a vector space model; and

determining an item recommendation by relating respective sets of item keywords with the expanded set of keywords and the initial set of extracted user profile keywords.

2. The computer implemented method as in claim 1 wherein the vector space model that is a topic vector space model (TVSM) that is arranged to produce relevancy vector space of commonly co-occurring keywords in the multitude of user profiles.

3. The computer implemented method as in claim 1 wherein the item recommendation is based on a query for one or more of:

keyword recommendations, movie recommendations, tv show recommendations, music recommendations, advertisement recommendations, or brand recommendations related to the initial set of extracted user profile keywords.

4. The computer implemented method as in claim 1 wherein the configuration of the data expansion model includes computing the frequency with which a keyword appears in conjunction with another keyword in one or more of the multitude of user profiles.

5. The computer implemented method as in claim 1 wherein computing the frequency with which a keyword appears in conjunction with another keyword in one or more of the multitude of user profiles for the data expansion model further includes:

measuring the degree to which the two keywords tend to occur together;

determining a ratio indicating the frequency with which the two keywords appear together; and

determining a correlation index indicating the likelihood that users interested in one of the keywords will be interested in the other keyword, as compared to an average user profile.

6. The computer implemented method as in claim 5 wherein processing a plurality of user social networking profiles to identify coincident keywords further includes:

processing the measured degree, the determined ratio and the correlation index to determine a percentage of co-occurrence for each of the keywords; and

determining, using the percentage of co-occurrence, a correlation ratio indicating how often a coincident keyword is present when another coincident keyword is present.

7. The computer implemented method as in claim 1 wherein the additional recommended keywords are determined using one or more of the identified co-occurring keywords from the data expansion model by using a term frequency—

inverse document frequency (idf) weighting calculation to determine the value of the extracted keyword from the subject user profile as an indication of user interest.

8. The computer implemented method as in claim 7 wherein weighing the importance of an extracted keyword from the subject user profile further includes treating the extracted keyword from the subject user profile and co-occurring keywords as nodes in an interconnected system, where the weights between nodes correspond to the strength of a statistical relation between the one or more extracted keywords and the coincident keywords.

9. The computer implemented method as in claim 1 wherein the search for recommendations for the subject user profile is configured to eliminate null results by comparing the set of extracted user profile with co-occurring keywords identified in a stored data expansion model that is configured to provide the measured degree to which at least two keywords tend to occur together in a multitude of user profiles.

10. The computer implemented method as in claim 1 wherein a respective set of the item related keywords is expanded according to the data expansion model, which is used to provide the measured degree to which at least two item related keywords tend to occur together in the multitude of user profiles.

11. A computer program product having computer-readable instructions stored on a non-transitory computer readable medium responsive to execution by a computing device that causes the computing device to perform operations comprising:

providing an item recommendation for a subject user profile, where each item has a respective set of associated keywords, the item recommendation being provided by;

processing the subject user profile to extract an initial set of keywords;

comparing the initial set of extracted user profile keywords with co-occurring keywords identified in a stored data expansion model that is configured to provide the measured degree to which at least two keywords tend to occur together in a multitude of user profiles;

expanding the initial set of extracted user profile keywords with additional related keywords, where the additional related keywords are determined using one or more of the identified co-occurring keywords from the expansion model that frequently appear together with the extracted user profile keywords in at least a portion of the multitude of user profiles, the expansion model being a vector space model; and

determining an item recommendation by relating respective sets of item keywords with the expanded set of keywords and the initial set of extracted user profile keywords.

12. The computer program product as in claim 11 wherein the vector space model that is a topic vector space model (TVSM) that is arranged to produce relevancy vector space of commonly co-occurring keywords in the multitude of user profiles.

13. The computer program product as in claim 11 wherein the item recommendation is a based on a query for one or more of:

keyword recommendations, movie recommendations, tv show recommendations, music recommendations, advertisement recommendations, or brand recommendations related to extracted keywords from the user profile.

14. The computer program product as in claim 11 wherein the configuration of the data expansion model includes computing the frequency with which a keyword appears in conjunction with another keyword in one or more of the multitude of user profiles.

15. The computer program product as in claim 11 wherein computing the frequency with which a keyword appears in conjunction with another keyword in one or more of the multitude of user profiles for the data expansion model further includes:

measuring the degree to which the two keywords tend to occur together;

determining a ratio indicating the frequency with which the two keywords appear together; and

determining a correlation index indicating the likelihood that users interested in one of the keywords will be interested in the other keyword, as compared to an average user profile.

16. The computer program product as in claim 15 wherein processing a plurality of user social networking profiles to identify coincident keywords further includes:

processing the measured degree, the determined ratio and the correlation index to determine a percentage of co-occurrence for each of the keywords; and

determining, using the percentage of co-occurrence, a correlation ratio indicating how often a coincident keyword is present when another coincident keyword is present.

17. The computer program product as in claim 11 wherein the additional recommended keywords are determined using one or more of the identified co-occurring keywords from the data expansion model by using a term frequency—

inverse document frequency (idf) weighting calculation to determine the value of the extracted keyword from the subject user profile as an indication of user interest.

18. The computer program product as in claim 17 wherein weighing the importance of an extracted keyword from the subject user profile further includes treating the extracted keyword from the subject user profile and co-occurring keywords as nodes in an interconnected system, where the weights between nodes correspond to the strength of a statistical relation between the one or more extracted keywords and the coincident keywords.

19. The computer program product as in claim 11 wherein the search for recommendations for the subject user profile is configured to eliminate null results by comparing the extracted keywords from the subject profile with co-occurring keywords identified in a stored data expansion model that is configured to provide the measured degree to which at least two keywords tend to occur together in a multitude of user profiles.

20. The computer program product as in claim 11 wherein a respective set of the item related keywords is expanded according to the data expansion model, the second expansion model being configured to provide the measured degree to which at least two item related keywords tend to occur together in a multitude of user profiles.

21. A data processing system including one or more computer processors configured to recommend products and services, the system comprising:

a data expansion model storing measurements regarding frequencies in which keywords tend to occur together in a multitude of user profiles;

a recommendation engine, in communication with the data expansion model;

configured to provide an item recommendation for a subject user profile, where a respective item has a respective set of associated keywords, the item recommendation being determined by;

processing the subject user profile to extract an initial set of keywords;

comparing the initial set of extracted user profile keywords with co-occurring keywords identified in the data expansion model to provide the measured degree to which at least two keywords tend to occur together in the multitude of user profiles;

expanding the initial set of extracted user profile keywords with additional related keywords, where the additional related keywords are determined using one or more of the identified co-occurring keywords from the expansion model that frequently appear together with the extracted user profile keywords in at least a portion of the multitude of user profiles, the expansion model being a vector space model; and

determining an item recommendation by relating respective sets of item keywords with the expanded set of keywords and the initial set of extracted user profile keywords.

22. A data processing system including one or more computer processors configured to recommend products and services, the system comprising:

a data expansion model storing measurements regarding frequencies in which keywords tend to occur together in a multitude of user profiles;

a recommendation engine, in communication with the data expansion model,

configured to provide an item recommendation for a subject user profile by;

receiving an initial set of keywords from a subject user profile;

comparing the initial set of extracted user profile keywords with co-occurring keywords identified in the data expansion model;

expanding the initial set of extracted user profile keywords with additional related keywords from the data expansion model, where the additional related keywords are determined using one or more of the identified co-occurring keywords from the expansion model that frequently appear together with the extracted user profile keywords in at least a portion of the multitude of user profiles in the data expansion model;

expanding keywords associated with items for recommendation using the data expansion model by determining additional item related keywords using one or more of the identified co-occurring keywords from the expansion model that frequently appear together with the extracted user profile keywords in at least a portion of the multitude of user profiles in the data expansion model, the expansion model being a vector space model;

determining an item recommendation by relating respective expanded sets of item keywords with the expanded set of keywords for the user profile and the initial set of extracted user profile keywords.