Search term prediction

• US 10,643,142 B2
• Filed: 08/02/2016
• Issued: 05/05/2020
• Est. Priority Date: 08/02/2016
• Status: Active Grant

• Current Assignee: Target Brands Inc. (Target Corporation)
• Original Assignee: Target Brands Inc. (Target Corporation)
• Inventors: Wan, Howard, Kuanr, Priyabrata, Bansal, Nikhil
• Primary Examiner(s): Chbouki, Tarek
• Assistant Examiner(s): Algibhah, Maher N
• Application Number:
  US15/226,187
• Publication Number:
  US 20180039699A1
• Time in Patent Office: 1,372 Days
• CPC Class Codes: G06F 16/3322 using system suggestions G06F16/3325 takes precedence
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

A system and computer-implemented method for predicting a search term of a search are disclosed. The method includes building, by a server, a search prediction dictionary database, the search prediction dictionary database including a plurality of search terms and a plurality of prediction scores corresponding to the plurality of search terms; receiving, by the server, one or more characters from a user device over a network, the one or more characters corresponding to the search; identifying, by the server, one or more predicted search terms based on the one or more characters, the identifying including determining one or more of the plurality of search terms including the one or more characters; and providing, by the server, a search term prediction list including the one or more predicted search terms to the user device for display on a display device of the user device.

Claims

1. A computer-implemented method for predicting a search term of a search in a retail environment, the method comprising:

building, by a server, a search prediction dictionary database, the search prediction dictionary database including a plurality of search terms and a plurality of prediction scores corresponding to the plurality of search terms, the building including;

retrieving click and sales data including search history data from a clickstream database, wherein the clickstream database is updated automatically such that older click and sales data that is older than a selected period of time is removed from the clickstream database and newer click and sales data that is within the selected period of time remains in the clickstream database;

creating the plurality of prediction scores using the search history data retrieved for the plurality of search terms, wherein the plurality of prediction scores are determined for each of the plurality of search terms based on a set of actions taken as a result of searches for the respective search term, the set of actions including a number of times the respective search term was searched, a click action in search results returned for the respective search term, an add to shopping cart action for an item in the search results returned for the respective search term, and a purchase action for an item in the search results returned for the respective term, wherein the purchase action is weighted greater than the add to shopping cart action, the add to shopping cart action is weighted greater than the click action, and the click action is weighted greater than the number of times the respective search term was searched, and actions from the set of actions taken more recently are given a greater weight than action from the set of actions taken less recently;

removing one or more searches in the search history data including special characters from the search history data; and

ordering the plurality of search terms within the search prediction dictionary database according to the plurality of prediction scores;

identifying, by the server, one or more predicted search terms based on the one or more characters, the identifying including determining one or more of the plurality of search terms including the one or more characters; and

providing, by the server, a search term prediction list including the one or more predicted search terms to the user device for display on a display device of the user device.

2. The computer-implemented method according to claim 1, wherein providing the search term prediction list includes sending the search term prediction list to the user device over a network.

3. The computer-implemented method according to claim 1, further comprising providing the search term prediction list when the search term prediction list includes more than a threshold number of search terms.

4. The computer-implemented method according to claim 1, further comprising correcting a spelling of the one or more characters when more than one character are received.

5. The computer-implemented method according to claim 1, wherein identifying the one or more predicted search terms includes using one or more previously submitted search terms identified from context information associated with the user device instead of the one or more search terms in the search prediction dictionary database.

6. The computer-implemented method according to claim 1, wherein identifying the one or more predicted search terms includes identifying search terms including the one or more characters, wherein the search terms include the one or more characters in a location other than a beginning of the one or more search terms.

7. The computer-implemented method according to claim 1, wherein the plurality of prediction scores are further determined based on one or more of a browsing history of the user device, one or more queries in the current browsing session of the user device, or one or more queries in previous browsing sessions of the user device.

8. The computer-implemented method according to claim 1, wherein the plurality of prediction scores are further determined based on a length of the search term compared to a length of a subset of characters of the search term.

9. The computer-implemented method according to claim 1, wherein the one or more predicted search terms are selected based on matching the one or more characters regardless of location within the one or more predicted search terms.

10. The computer-implemented method according to claim 1, wherein each of the plurality of prediction scores are further weighted based on a comparison between a length of the one or more characters provided by the user device and a length of a subset of characters of a predicted search term.

11. A system for predicting a search term in a retail environment, comprising:

a server that includes;

one or more processors; and

a memory storing instructions that, when executed by the one or more processors, cause the system to perform;

a search prediction dictionary builder that generates a search prediction dictionary database that stores one or more keywords each associated with one or more prediction scores, including;

retrieving click and sales data including search history data from a clickstream database, wherein the clickstream database is updated automatically such that older click and sales data that is older than a selected period of time is removed from the clickstream database and newer click and sales data that is within the selected period of time remains in the clickstream database;

creating the plurality of prediction scores using the search history data retrieved for the plurality of search terms, wherein the plurality of prediction scores are determined for each of the plurality of search terms based on a set of actions taken as a result of the searches for the respective search term, the set of actions including a number of times the respective search term was searched, a click action in search results returned for the respective search term, an add to shopping cart action for an item in the search results returned for the respective search term, and a purchase action for an item in the search results returned for the respective term, wherein the purchase action is weighted greater than the add to shopping cart action, the add to shopping cart action is weighted greater than the click action, and the click action is weighted greater than the number of times the respective search term was searched, and actions from the set of actions taken more recently are given a greater weight than action from the set of actions taken less recently;

removing one or more searches in the search history data including special characters from the search history data; and

ordering the plurality of search terms within the search prediction dictionary database according to the plurality of prediction scores;

the search term predictor that receives the at least one character provides a search term prediction list that includes the one or more predicted keywords.

12. The system according to claim 11, wherein the server is connectable to a user device over a network, the user device including a display device and an input device, and the server can receive the at least one character from the user device.

13. The system according to claim 11, wherein the server is connectable to a user device over a network and the server can send the search term prediction list to the user device for display on a display device of the user device.

14. The system according to claim 11, wherein the one or more predicted keywords include the at least one character in a location other than a beginning of the one or more keywords in the search prediction dictionary database.

15. The system according to claim 11, wherein the plurality of prediction scores are further determined based on one or more of a browsing history of the user device, one or more queries in the current browsing session of the user device, or one or more queries in previous browsing sessions of the user device.

16. The system according to claim 11, wherein the plurality of prediction scores are further determined based on a length of the search term compared to a length of a subset of characters of the search term.

17. The system according to claim 11, wherein the one or more predicted search terms are selected based on matching the one or more characters regardless of location within the one or more predicted search terms.

18. The system according to claim 11, wherein each of the plurality of prediction scores are further weighted based on a comparison between a length of the one or more characters provided by the user device and a length of a subset of characters of a predicted search term.