Ranking search results based on color

• US 9,563,667 B2
• Filed: 05/24/2016
• Issued: 02/07/2017
• Est. Priority Date: 12/28/2012
• Status: Active Grant

• Current Assignee: Walmart Apollo, LLC (WalMart Inc.)
• Original Assignee: Wal-Mart Stores Texas LLC (WalMart Inc.)
• Inventors: Vuong, Ba-Quy, Gattani, Abhishek, Tholiya, Namrata Pramodkumar
• Primary Examiner(s): Morrison, Jay
• Assistant Examiner(s): Hoang, Ken
• Application Number:
  US15/163,554
• Publication Number:
  US 20160267088A1
• Time in Patent Office: 259 Days
• Field of Search: 715/738, 715/753, 715/733, 715/226, 715/234, 715/253, 715/719, 715/744, 715/746, 715/747, 707/769, 707/E17.014, 707/722, 707/E17.03, 707/E17.044, 707/706, 707/708, 707/710, 707/731, 707/732, 707/737, 707/753, 707/812, 709/203, 709/206, 709/213, 705/4.43, 705/12, 705/14.23, 705/14.49, 705/14.53, 705/2, 705/26.1, 705/26.5, 705/26.63, 705/30, 705/7.11, 705/7.36, 701/1, 701/533, 704/E11.001, 704/2, 704/270, 704/8, 704/9
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/248 Presentation of query results
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 30/0641 Shopping interfaces

Abstract

Ranking search results based on color including evaluating whether a color term in a search query was intended to identify an actual color and returning different search results based on that determination. When an actual color was intended, an initial product ranking is updated based on color similarity of the product. In some embodiments, a color similarity score is calculated using a color difference formula, a cosine similarity measurement is determined of a cosine of an angle between a query vector and a title vector, and the color is considered to have been intended to identify the actual color when the cosine similarity measurement is less than a threshold value. In particular embodiments, the query vector includes individual terms of the search query and a number of occurrences of each individual term. Different embodiments include computer implemented systems and methods.

Claims

1. A method comprising:

receiving, by one or more processors of a server, an initial search query from a user device over a communication network;

identifying, by the one or more processors of the server, a reference to a first color in the initial search query;

determining, by the one or more processors of the server, whether the reference to the first color is intended to identify any actual color;

in response to determining that the reference to the first color is not intended to identify any actual color, the one or more processors of the server performing operations comprising;

executing the initial search query to obtain a search result;

ranking the search result without regard to the reference to the first color; and

transmitting a first displayable result of the ranking of the search result to the user device over the communication network; and

in response to determining that the reference to the first color is intended to identify any actual color, the one or more processors of the server performing operations comprising;

identifying a plurality of products based on the initial search query;

ranking the plurality of products;

identifying one or more second colors associated with the plurality of products;

for each second color of the one or more second colors, determining a color similarity score between the each second color and the first color wherein the color similarity score is calculated using a color difference formula;

generating an updated ranking of the plurality of products based on the one or more second colors associated with the plurality of products and the color similarity score of each of the one or more second colors; and

transmitting a second displayable result of the updated ranking of the plurality of products to the user device over the communication network.

2. The method of claim 1 further comprising, after the determining that the reference to the first color is intended to identify any actual color:

determining that a respective second color associated with a particular product of the plurality of products is identical or similar to the first color;

increasing a ranking score associated with the particular product; and

generating the updated ranking of the plurality of products based on the increased ranking score associated with the particular product.

3. The method of claim 2 wherein the determining, by the one or more processors of the server, whether the reference to the first color is intended to identify any actual color, comprises:

parsing the initial search query as a query vector;

deriving a cosine similarity measurement of a cosine of an angle between the query vector and a title vector of an identified title; and

determining that the first color is intended to identify the actual color when the cosine similarity measurement is less than a threshold value.

4. The method of claim 2, wherein the increasing the ranking score associated with the particular product comprises:

increasing the ranking score associated with the particular product by a first amount when the respective second color associated with the particular product is identical to the first color; and

increasing the ranking score associated with the particular product by a second amount when the respective second color associated with the particular product is similar to the first color,

wherein the second amount is less than the first amount.

5. The method of claim 1 wherein the determining, by the one or more processors of the server, whether the reference to the first color is intended to identify any actual color, comprises:

parsing the initial search query as a query vector, the query vector comprising individual terms of the initial search query and, for each of the individual terms of the initial search query, a number of occurrences of the individual term of the initial search query;

deriving a cosine similarity measurement of a cosine of an angle between the query vector and a title vector derived from individual titles of search results associated with the initial search query; and

determining that the first color is intended to identify the actual color when the cosine similarity measurement is less than a threshold value.

6. The method of claim 1 wherein the identifying the reference to the first color in the initial search query includes comparing each term in the initial search query with a listing of known color terms.

7. The method of claim 1 wherein the ranking comprises ranking with a BM25F ranking function.

8. The method of claim 1 further comprising creating a second search query by removing color-related terms from the initial search query.

9. The method of claim 8, wherein the color-related terms modify or characterize a color.

10. A method comprising:

receiving, by one or more processors of a server, an initial search query from a user device over a communication network;

identifying, by the one or more processors of the server, a reference to a first color in the initial search query;

determining, by the one or more processors of the server, whether the reference to the first color is intended to identify any actual color;

in response to determining that the reference to the first color is not intended to identify an actual color, the one or more processors of the server performing operations comprising;

executing the initial search query to obtain a search result;

ranking the search result without regard to the reference to the first color; and

transmitting a first displayable result of the ranking of the search result to the user device over the communication network; and

in response to determining that the reference to the first color is intended to identify the actual color, the one or more processors of the server performing operations comprising;

creating a rewritten search query which is different from the initial search query by re-writing the initial search query;

executing the rewritten search query to identify a plurality of products based on the rewritten search query;

determining one or more products of the plurality of products as being associated with one or more colors identical or similar to the first color;

transforming the ranking of the search result comprising;

increasing a ranking score associated with each of the one or more products based on similarity between the first color and the one or more colors associated with the each of the one or more products;

ranking the plurality of products in the ranking of the search result based on the initial search query and the ranking score associated with each of the one or more products; and

transmitting a second displayable result of the ranking of the plurality of products to the user device over the communication network.

11. The method of claim 10 wherein the increasing the ranking score associated with each of the one or more products comprises:

increasing the ranking score associated with the each of the one or more products by a first amount in an event that the one or more colors associated with the each of the one or more products is identical to the first color; and

increasing the ranking score associated with the each of the one or more products by a second amount in an event that the one or more colors associated with the each of the one or more products is similar to the first color,

wherein the second amount is less than the first amount.

12. The method of claim 10 wherein the determining one or more products of the plurality of products as being associated with one or more colors identical or similar to the first color is based at least in part on a color similarity score determined by a color difference formula, the color similarity score being based on a similarity between the first color and the one or more colors associated with the each of the one or more products.

13. The method of claim 10 wherein the determining, by the one or more processors of the server, whether the reference to the first color is intended to identify any actual color, comprises:

parsing the initial search query as a query vector, the query vector comprising individual terms of the initial search query and, for each of the individual terms of the initial search query, a number of occurrences of the individual term of the initial search query;

deriving a cosine similarity measurement of a cosine of an angle between the query vector and a title vector derived from individual titles of search results associated with the initial search query; and

determining that the first color is intended to identify the actual color when the cosine similarity measurement is less than a threshold value.

14. The method of claim 10 wherein the identifying the reference to the first color in the initial search query comprises comparing each term in the initial search query with a listing of known color terms.

15. The method of claim 10 wherein the increasing the ranking score associated with the each of the one or more products comprises increasing the ranking score by a value that is proportional to a color similarity between the first color and the one or more colors associated with the each of the one or more products.

16. The method of claim 10 wherein the creating the rewritten search query comprises creating the rewritten search query by removing each reference to any color in the initial search query.

17. A system comprising:

a database storing data associated with a plurality of products; and

a server communicatively coupled to the database that receives an initial search query from a user device over a communication network, wherein, in response to receiving the initial search query, the server;

identifies a reference to a first color in the initial search query;

determines whether the reference to the first color is intended to identify any actual color;

in response to determining that the reference to the first color is not intended to identify any actual color, performs operations comprising;

executing the initial search query by accessing the database to obtain a search result;

ranking the search result without regard to the reference to the first color; and

transmitting a first displayable result of the ranking of the search result to the user device over the communication network; and

in response to determining that the reference to the first color is intended to identify an actual color, performs operations comprising;

identifying a plurality of products based on the initial search query by accessing the database;

ranking the plurality of products;

identifying one or more second colors associated with the plurality of products;

for each second color of the one or more second colors, determining a color similarity score between the each second color and the first color;

generating an updated ranking of the plurality of products based on the one or more second colors associated with the plurality of products and the color similarity score of each of the one or more second colors; and

transmitting a second displayable result of the updated ranking of the plurality of products to the user device over the communication network.

18. The system of claim 17, wherein the server further performs operations comprising:

determining that a respective second color associated with a particular product of the plurality of products is identical or similar to the first color;

increasing a ranking score associated with the particular product by a first amount in an event that the respective second color associated with the particular product is identical to the first color;

increasing the ranking score associated with the particular product by a second amount in an event that the respective second color associated with the particular product is similar to the first color, wherein the second amount less than the first amount; and

generating a ranking of the plurality of products based on the ranking score associated with the particular product.

19. The system of claim 17, wherein the server further performs operations comprising:

parsing the initial search query as a query vector, the query vector comprising individual terms of the initial search query and, for each of the individual terms of the initial search query, a number of occurrences of the individual term of the initial search query;

deriving a cosine similarity measurement of a cosine of an angle between the query vector and a title vector derived from individual titles of search results associated with the initial search query; and

determining that the first color is intended to identify the actual color when the cosine similarity measurement is less than a threshold value.

20. The system of claim 17, wherein the server further performs operations comprising:

determining one or more products of the plurality of products as being associated with one or more colors identical or similar to the first color based at least in part on a color similarity score determined by a color difference formula, the color similarity score being based on a similarity between the first color and the one or more colors associated with each of the one or more products;

transforming the ranking of the search result comprising;

increasing a ranking score associated with each of the one or more products based on the color similarity score, comprising;

increasing the ranking score associated with the each of the one or more products by a first amount in an event that the one or more colors associated with the each of the one or more products is identical to the first color; and

increasing the ranking score associated with the each of the one or more products by a second amount that is proportional to a color similarity between the first color and the one or more colors associated with the each of the one or more products in an event that the one or more colors associated with the each of the one or more products is similar to the first color, wherein the second amount is less than the first amount; and

ranking the plurality of products in the search result ranking based on the initial search query and the ranking score associated with each product in the plurality of products; and

transmitting a displayable result of the ranking of the search result of the plurality of products to the user device over the communication network.