Recommending keywords

• US 9,117,006 B2
• Filed: 04/19/2013
• Issued: 08/25/2015
• Est. Priority Date: 04/25/2012
• Status: Active Grant

• Current Assignee: Alibaba Group Holding Ltd.
• Original Assignee: Alibaba Group Holding Ltd.
• Inventors: Zhu, Daoyong, Dong, Fangying
• Primary Examiner(s): Stevens, Robert
• Application Number:
  US13/866,953
• Publication Number:
  US 20130290320A1
• Time in Patent Office: 858 Days
• Field of Search: 707/723
• US Class Current: 1/1
• CPC Class Codes: G06F 16/90335 Query processing
• CPC Class Codes: G06F 16/95 Retrieval from the web

Abstract

Recommending keywords is disclosed, including: receiving a set of product information including a product title; extracting and parsing the product title into a set of parsed elements; finding a plurality of candidate keywords corresponding to at least a subset of the set of parsed elements based at least in part on stored mappings between parsed data and keywords; determining a plurality of composite correlation scores for corresponding ones of the plurality of candidate keywords; sorting at least a subset of the plurality of candidate keywords into a ranked list based on at least a subset of the plurality of composite correlation scores; and selecting a set of one or more keywords to recommend from the plurality of candidate keywords based at least in part on the ranked list.

Claims

1. A system for recommending keywords, comprising:

one or more processors configured to;

receive a set of product information including a product title;

extract and parse the product title into a set of parsed elements;

find a plurality of candidate keywords corresponding to at least a subset of the set of parsed elements based at least in part on stored predetermined mappings between parsed data and keywords;

determine a plurality of composite correlation scores for corresponding ones of the plurality of candidate keywords, wherein to determine the plurality of composite correlation scores includes to determine a first composite correlation score associated with a first candidate keyword of the plurality of candidate keywords, wherein to determine the first composite correlation score associated with the first candidate keyword includes to determine an industry index value associated with the first candidate keyword, including to;

determine a first similarity value between one or more industries associated with the first candidate keyword and one or more industries associated with sets of product information that are relevant to the first candidate keyword;

determine a second similarity value between the one or more industries associated with the first candidate keyword and one or more industries of one or more seller users associated with the sets of product information that are relevant to the first candidate keyword; and

determine the industry index value associated with the first candidate keyword based at least in part on a combination of the first similarity value and the second similarity value; and

sort at least a subset of the plurality of candidate keywords into a ranked list based on at least a subset of the plurality of composite correlation scores; and

select a set of one or more keywords to recommend from the plurality of candidate keywords based at least in part on the ranked list; and

one or more memories coupled to the one or more processors and configured to provide the one or more processors with instructions.

2. The system of claim 1, wherein the one or more processors are configured to determine the at least subset of the set of parsed elements, including to:

determine a correlation value between each of the set of parsed elements and the set of product information; and

determine the at least subset of the set of parsed elements to include M parsed elements of the set of parsed elements associated with the highest respective correlation values, wherein M comprises a positive integer.

3. The system of claim 1, wherein at least a subset of the keywords included in the predetermined mappings between parsed data and keywords is determined from historical search logs.

4. The system of claim 1, wherein to determine the plurality of composite correlation scores includes to determine the first composite correlation score associated with the first candidate keyword based at least in part on a text-based correlation value between the first candidate keyword and a corresponding parsed element from the set of parsed elements.

5. The system of claim 1, wherein to determine the plurality of composite correlation scores includes to determine the first composite correlation score associated with the first candidate keyword based at least in part on a Gini index value associated with the first candidate keyword, wherein the Gini index value is used to reflect a probability that sets of product information that match the first candidate keyword are displayed among search results.

6. The system of claim 1, wherein to determine the plurality of composite correlation scores includes to determine the first composite correlation score associated with the first candidate keyword based at least in part on a supply and demand index value associated with the first candidate keyword, wherein the supply and demand index value is used to reflect a supply and demand relationship relevant to the first candidate keyword.

7. The system of claim 1, wherein to determine the plurality of composite correlation scores includes to determine the first composite correlation score associated with the first candidate keyword based at least on a combination of a text-based correlation value between the first candidate keyword and a corresponding parsed element from the set of parsed elements, a Gini index value associated with the first candidate keyword, a supply and demand index value associated with the first candidate keyword, and the industry index value associated with the first candidate keyword.

8. The system of claim 1, wherein the one or more processors are further configured to clean the product title prior to parsing the product title into the set of parsed elements.

9. The system of claim 1, wherein the one or more processors are further configured to receive an update to the set of product information, the update being based at least in part on the set of one or more keywords.

10. The system of claim 1, wherein to find the plurality of candidate keywords corresponding to the at least subset of the set of parsed elements comprises to:

identify a matching piece of parsed data of the stored predetermined mappings;

determine one or more keywords that correspond to the matching piece of parsed data in the stored predetermined mappings; and

include the one or more keywords that correspond to the matching piece of parsed data in the plurality of candidate keywords.

11. A method for recommending keywords, comprising:

receiving a set of product information including a product title;

extracting and parsing the product title into a set of parsed elements;

finding a plurality of candidate keywords corresponding to at least a subset of the set of parsed elements based at least in part on stored predetermined mappings between parsed data and keywords;

determining a plurality of composite correlation scores for corresponding ones of the plurality of candidate keywords, wherein determining the plurality of composite correlation scores includes determining a first composite correlation score associated with a first candidate keyword of the plurality of candidate keywords, wherein determining the first composite correlation score associated with the first candidate keyword includes determining an industry index value associated with the first candidate keyword, including;

determining a first similarity value between one or more industries associated with the first candidate keyword and one or more industries associated with sets of product information that are relevant to the first candidate keyword;

determining a second similarity value between the one or more industries associated with the first candidate keyword and one or more industries of one or more seller users associated with the sets of product information that are relevant to the first candidate keyword; and

determining the industry index value associated with the first candidate keyword based at least in part on a combination of the first similarity value and the second similarity value; and

sorting at least a subset of the plurality of candidate keywords into a ranked list based on at least a subset of the plurality of composite correlation scores; and

selecting a set of one or more keywords to recommend from the plurality of candidate keywords based at least in part on the ranked list.

12. The method of claim 11, wherein the at least subset of the set of parsed elements is determined based on:

determining a correlation value between each of the set of parsed elements and the set of product information; and

determining the at least subset of the set of parsed elements to include M parsed elements of the set of parsed elements associated with the highest respective correlation values, wherein M comprises a positive integer.

13. The method of claim 11, wherein at least a subset of the keywords included in the predetermined mappings between parsed data and keywords is determined from historical search logs.

14. The method of claim 11, wherein determining the plurality of composite correlation scores includes determining the first composite correlation score associated with the first candidate keyword based at least in part on a text-based correlation value between the first candidate keyword and a corresponding parsed element from the set of parsed elements.

15. The method of claim 11, wherein determining the plurality of composite correlation scores includes determining the first composite correlation score associated with the first candidate keyword based at least in part on a Gini index value associated with the first candidate keyword, wherein the Gini index value is used to reflect a probability that sets of product information that match the first candidate keyword are displayed among search results.

16. The method of claim 11, wherein determining the plurality of composite correlation scores includes determining the first composite correlation score associated with the a first candidate keyword based at least in part on a supply and demand index value associated with the first candidate keyword, wherein the supply and demand index value is used to reflect a supply and demand relationship relevant to the first candidate keyword.

17. The method of claim 11, wherein determining the plurality of composite correlation scores includes determining the first composite correlation score associated with the first candidate keyword based at least on a combination of a text-based correlation value between the first candidate keyword and a corresponding parsed element from the set of parsed elements, a Gini index value associated with the first candidate keyword, a supply and demand index value associated with the first candidate keyword, and the industry index value associated with the first candidate keyword.

18. The method of claim 11, further comprising cleaning the product title prior to parsing the product title into the set of parsed elements.

19. A computer program product for recommending keywords, the computer program product being embodied in a non-transitory computer readable storage medium and comprising computer instructions for:

receiving a set of product information including a product title;

extracting and parsing the product title into a set of parsed elements;

finding a plurality of candidate keywords corresponding to at least a subset of the set of parsed elements based at least in part on stored predetermined mappings between parsed data and keywords;

determining a plurality of composite correlation scores for corresponding ones of the plurality of candidate keywords, wherein determining the plurality of composite correlation scores includes determining a first composite correlation score associated with a first candidate keyword of the plurality of candidate keywords, wherein determining the first composite correlation score associated with the first candidate keyword includes determining an industry index value associated with the first candidate keyword, including;

determining a first similarity value between one or more industries associated with the first candidate keyword and one or more industries associated with sets of product information that are relevant to the first candidate keyword;

determining a second similarity value between the one or more industries associated with the first candidate keyword and one or more industries of one or more seller users associated with the sets of product information that are relevant to the first candidate keyword; and

determining the industry index value associated with the first candidate keyword based at least in part on a combination of the first similarity value and the second similarity value; and

sorting at least a subset of the plurality of candidate keywords into a ranked list based on at least a subset of the plurality of composite correlation scores; and

selecting a set of one or more keywords to recommend from the plurality of candidate keywords based at least in part on the ranked list.