Image search using intersected predicted queries

• US 10,552,478 B1
• Filed: 12/28/2016
• Issued: 02/04/2020
• Est. Priority Date: 12/28/2016
• Status: Active Grant

• Current Assignee: Shutterstock Incorporated
• Original Assignee: Shutterstock Incorporated
• Inventors: Lev-Tov, Manor, Hurst, Nathan
• Primary Examiner(s): Alam, Shahid A
• Application Number:
  US15/393,016
• Time in Patent Office: 1,133 Days
• Field of Search: 707706
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/58 Retrieval characterised by using metadata, e.g. metadata not derived from the content or metadata generated manually
• CPC Class Codes: G06F 16/5838 using colour
• CPC Class Codes: G06F 16/5866 using information manually generated, e.g. tags, keywords, comments, manually generated location and time information
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation

Abstract

A method for receiving a first user query from a user for searching an item, forming a first filter based on the first user query, and forming a first filtered item collection is provided. The method includes predicting a new query based on the first user query and a historical query log, forming a second filter for the new query, and applying the second filter to the first filtered item collection to form a second filtered item collection. Further, associating an item score to each of a plurality of items in the first and second filtered item collections, sorting the plurality of items in the first and second filtered item collections according to the item score associated to each of the plurality of items, and providing, to a user display, an item in the plurality of items in the first or second filtered item collections according to a sorting order.

Claims

1. A computer-implemented method, comprising:

receiving a first user query from a user for searching for an item in an item database;

forming a first filter based on the first user query;

forming a first filtered item collection comprising a plurality of items from the item database;

forming a new query with the first user query and at least one term from a historical query log;

re-ordering a plurality of terms in the new query according to an expected scope of search results for a permutated new query;

forming a second filter for the new query;

applying the second filter to the first filtered item collection to form a second filtered item collection;

associating an item score to each of a plurality of items in the first filtered item collection and in the second filtered item collection;

sorting the plurality of items in the first filtered item collection and in the second filtered item collection according to the item score associated to each of the plurality of items; and

providing, for display to a user in a results panel, at least one item from the plurality of items in the first filtered item collection and in the second filtered item collection.

2. The method of claim 1, wherein forming the first filtered item collection comprises applying the first filter to a plurality of image files stored in an image database.

3. The method of claim 1, wherein predicting a new query based on the first user query and a historical query log comprises matching at least a portion of the new query with at least a portion of the first user query and a portion of a query from the historical query log.

4. The method of claim 1, wherein predicting a new query based on the first user query and the historical query log comprises:

forming a third filter for the new query; and

forming a third filtered item collection by applying the third filter to the first filtered item collection, wherein the third filtered item collection and the second filtered item collection do not overlap.

5. The method of claim 1, wherein forming the second filtered item collection comprises applying the second filter to the first filtered item collection.

6. The method of claim 1, wherein associating an item score to one of a plurality of items in the first filtered item collection and a plurality of items in the second filtered item collection comprises associating a probability with the one of the plurality of items for which a user clicks on a thumbnail associated with the one of the plurality of items, based on an interaction history information and a relevance feedback information.

7. The method of claim 1, wherein associating an item score to each of a plurality of items in the first filtered item collection and in the second filtered item collection comprises finding a correlation between the first user query and interaction history information associated with the user.

8. The method of claim 1, wherein associating an item score to each of a plurality of items in the first filtered item collection and in the second filtered item collection comprises finding a correlation between the first user query and object recognition information associated with each of the plurality of items and associating a higher item score for a higher correlation between the first user query and object recognition information.

9. A system comprising:

one or more processors to;

receive a first user query from a user for searching for an item in an item database;

form a first filter based on the first user query;

form a first filtered item collection comprising a plurality of items from the item database;

determine a probability that a non-matching portion is associated with at least one term from the first user query for a predicted query string in a second filter;

sort the plurality of items in the first filtered item collection and in a second filtered item collection according to an interaction history information;

add the predicted query string to a plurality of predicted queries;

form a second filter for a new query;

apply the second filter to the first filtered item collection to form a second filtered item collection;

associate an item score to each of a plurality of items in the first filtered item collection and in the second filtered item collection;

sort the plurality of items in the first filtered item collection and in the second filtered item collection according to the item score associated to each of the plurality of items; and

provide, for display to a user in a results panel, at least one item from the plurality of items in the first filtered item collection and in the second filtered item collection.

10. The system of claim 9, wherein the one or more processors further select at least one term from the first user query, and to form the first filter based on the first user query, based on a predicted query string from a query stream log and the at least one term from the first user query.

11. The system of claim 9, wherein the one or more processors form an item search thread for the new query.

12. The system of claim 9, wherein to predict a new query based on the first user query and a historical query log, the one or more processors further re-order a plurality of terms in the new query according to an expected scope of search results for the new query.

13. The system of claim 9, wherein to predict a new query based on the first user query and a historical query log the one or more processors match at least a portion of the new query with at least a portion of the first user query and a portion of a query from the historical query log.

14. The system of claim 9, wherein the one or more processors further display a pull-down menu for a user that includes the new query.

15. The system of claim 9, wherein to predict a new query based on the first user query and a historical query log the one or more processors:

form the new query with the first user query and at least one term from the historical query log; and

re-order a plurality of terms in the new query according to an expected scope of search results for a permutated new query.

16. The system of claim 9, wherein to predict a new query based on the first user query and a historical query log the one or more processors:

form a third filter for the new query; and

form a third filtered item collection by applying the third filter to the first filtered item collection, wherein the third filtered item collection and the second filtered item collection do not overlap.

17. The system of claim 9, wherein to form the second filtered item collection the one or more processors apply the second filter to the first filtered item collection.

18. A non-transitory, computer-readable medium storing instructions which, when executed by a processor, cause a computer to perform a method, the method comprising:

receiving a first user query from a user for searching for an item in an item database;

forming a first filter based on the first user query;

forming a first filtered item collection comprising a plurality of items from the item database;

forming a new query with the first user query and at least one term from a historical query log;

re-ordering a plurality of terms in the new query according to an expected scope of search results for a permutated, new query;

forming a second filter for the new query;

applying the second filter to the first filtered item collection to form a second filtered item collection;

associating an item score to each of a plurality of items in the first filtered item collection and in the second filtered item collection;

sorting the plurality of items in the first filtered item collection and in the second filtered item collection according to the item score associated to each of the plurality of items; and

providing, for display to a user in a results panel, at least one item from the plurality of items in the first filtered item collection and in the second filtered item collection.

19. The non-transitory, computer-readable medium of claim 18, wherein in the method, forming the first filtered item collection comprises applying the first filter to a plurality of image files stored in an image database.

20. The non-transitory, computer-readable medium of claim 18, wherein in the method, predicting a new query based on the first user query and a historical query log comprises matching at least a portion of the new query with at least a portion of the first user query and a portion of a query from the historical query log.