Method and apparatus for sorting inquiry results

• US 8,504,563 B2
• Filed: 07/22/2011
• Issued: 08/06/2013
• Est. Priority Date: 07/26/2010
• Status: Expired due to Fees

• Current Assignee: Alibaba Group Holding Ltd.
• Original Assignee: Alibaba Group Holding Ltd.
• Inventors: Chen, Chao, Han, Xiaomei
• Primary Examiner(s): Jalil, Neveen Abel
• Assistant Examiner(s): KINSAUL, DANIEL W
• Application Number:
  US13/262,611
• Publication Number:
  US 20130132356A1
• Time in Patent Office: 746 Days
• Field of Search: None
• US Class Current: 707/728
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06Q 30/0282 Rating or review of business operators or products

Abstract

Sorting inquiry results includes, based on extracted inquiry results matching search conditions of a user, collecting features of the inquiry results. The collected features may be used as features of a respective inquiry result and feature fitting may be conducted based on a support vector machine (SVM) regression model to obtain a feature fitting value of the respective inquiry result. The inquiry results may be sorted based on relevancy values of the inquiry results, and, for inquiry results having a same relevancy level, the inquiry results may be sorted in a top-down manner based on feature fitting values of the inquiry results.

Claims

1. A method of sorting inquiry results in a search engine, performed by one or more processors configured with computer readable instructions, the method comprising:

based on extracted inquiry results matching search conditions of a user, collecting one or more features of a respective inquiry result, the one or more features of the inquiry results including at least a quality factor indicating a quality of the respective inquiry result and one or any combination of the following factors;

an active level factor indicating an active level of a supplier relating to the respective inquiry result, an industry factor indicating an industry of the supplier relating to the respective inquiry result, an information click conversion rate and an information feedback conversion rate of the supplier relating to the respective inquiry result, the collecting the quality factor including;

obtaining one or more quality indicators of the respective inquiry result according to statistics of information of a database, the one or more quality indicators including one or any combination of;

a completion rate of information including a ratio of a number of filled fields to a number of fillable fields of the respective inquiry result, a completion rate of attributes including a ratio of a number of filled attributes to a number of fillable attributes of the respective inquiry result, and a completion rate of detailed description including a pre-set value based on a number of words included in a description of the respective inquiry result; and

conducting a linear fitting of the one or more quality indicators to obtain the quality factor;

using collected factors as the one or more features of the respective inquiry result and conducting feature fitting based on a support vector machine (SVM) regression model to obtain a feature fitting value of the respective inquiry result; and

sorting the inquiry results based on relevancy values of the inquiry results, and, for inquiry results having a same or similar relevancy values, sorting the inquiry results in a top-down manner based on feature fitting values of the inquiry results.

2. The method as recited in claim 1, wherein the following factors further include a timeliness factor indicating a timeliness of the respective inquiry result.

3. The method as recited in claim 1, wherein the collecting the active level factor comprises:

obtaining one or more active level indicators of the supplier relating to the respective inquiry result, the one or more active level indicators including one or any combination of;

an online time of the supplier relating to the inquiry result and an active grade of the supplier relating to the inquiry result; and

conducting a linear fitting of the one or more active level indicators to obtain the active level factor.

4. The method as recited in claim 1, wherein the collecting the information click conversion rate comprises:

obtaining a number of times the information being exposed and a number of times the information being clicked in a period of time according to statistics of information of a user log; and

calculating a ratio of the number of times the information being exposed to the number of times the information being clicked as the information click conversion rate.

5. The method as recited in claim 1, wherein the collecting the feedback conversion rate of information comprises:

obtaining a number of times the information being clicked and a number of times the information with feedbacks in a period of time according to statistics of information of a user log; and

calculating a ratio of the number of times the information with feedbacks to the number of times the information being clicked as the information feedback conversion rate.

15. The method as recited in claim 1, wherein the one or more features further comprise a click feedback rate obtained by calculating a ratio of a number of times feedback is received to a number of times information is clicked.

6. An apparatus for sorting inquiry results in a search engine, comprising:

one or more processors; and

memory having stored therein modules that are executable on the one or more processors, the modules including;

a first collection module that, based on extracted inquiry results matching search conditions of the user, collects one or more features of a respective inquiry result, the one or more features of the inquiry results including at least a quality factor indicating a quality of a respective inquiry result and one or in any combination of the following factors;

an active level factor indicating an active level of a supplier relating to the respective inquiry result, an industry factor indicating an industry of the supplier relating to the respective inquiry result, an information click conversion rate and an information feedback conversion rate of the supplier relating to the respective inquiry result, the collecting the quality factor including;

obtaining one or more quality indicators of the respective inquiry result according to statistics of information of a database, the one or more quality indicators including one or any combination of;

a completion rate of information including a ratio of a number of filled fields to a number of fillable fields of the respective inquiry result, a completion rate of attributes including a ratio of a number of filled attributes to a number of fillable attributes of the respective inquiry result, and a completion rate of detailed description including a pre-set value based on a number of words included in a description of the respective inquiry result; and

conducting a linear fitting of the one or more quality indicators to obtain the quality factor;

a feature fitting module that uses the collected factors as one or more features of the respective inquiry result and conducts feature fitting based on a support vector machine (SVM) regression model to obtain a feature fitting value of the respective inquiry result; and

a sorting module that sorts the inquiry results based on relevancy values of the inquiry results, and, for inquiry results having a same or similar relevancy values, sorting the inquiry results in a top-down manner based on feature fitting values of the inquiry results.

7. The apparatus as recited in claim 6, further comprising a second collection module that collects a timeliness factor indicating a timeliness of the respective inquiry result.

16. The apparatus as recited in claim 6, wherein the one or more features further comprise a click feedback rate obtained by calculating a ratio of a number of times feedback is received to a number of times information is clicked.

17. The apparatus as recited in claim 6, wherein the first collection module collects the active level factor by:

obtaining one or more active level indicators of the supplier relating to the respective inquiry result, the one or more active level indicators including one or any combination of;

an online time of the supplier relating to the inquiry result and an active grade of the supplier relating to the inquiry result; and

conducting a linear fitting of the one or more active level indicators to obtain the active level factor.

8. One or more computer-readable storage media comprising computer-readable instructions executable by one or more processors that, when executed by the one or more processors, cause the one or more processors to perform operations comprising:

collecting one or more features of a respective inquiry result of inquiry results matching a search condition of a user, the one or more features of the inquiry results including at least a quality factor indicating quality of the respective inquiry result and one or any combination of the following factors;

an active level factor indicating an active level of a supplier relating to the respective inquiry result, an industry factor indicating an industry of the supplier relating to the respective inquiry result, an information click conversion rate and an information feedback conversion rate of the supplier relating to the respective inquiry result, the collecting the quality factor including;

obtaining one or more quality indicators of the respective inquiry result according to statistics of information of a database, the one or more quality indicators including one or any combination of;

a completion rate of information including a ratio of a number of filled fields to a number of fillable fields of the respective inquiry result, a completion rate of attributes including a ratio of a number of filled attributes to a number of fillable attributes of the respective inquiry result, and a completion rate of detailed description including a pre-set value based on a number of words included in a description of the respective inquiry result; and

conducting a linear fitting of the one or more quality indicators to obtain the quality factor

using collected factors as the one or more features of the respective inquiry result and conducting feature fitting based on a support vector machine (SVM) regression model to obtain a feature fitting value of the respective inquiry result; and

sorting the inquiry results based on relevancy values of the inquiry results, and, for inquiry results having a same or similar relevancy values, sorting the inquiry results in a top-down manner based on feature fitting values of the inquiry results.

9. The one or more computer-readable media as recited in claim 8, wherein the similar relevancy values are values whose differences are within a threshold.

10. The one or more computer-readable media as recited in claim 8, wherein the one or more active level factors comprise an online time of the supplier and an active grade of the supplier.

11. The one or more computer-readable media as recited in claim 8, wherein the industry factor indicates the industry of the supplier relating to the respective inquiry result according to user log information.

12. The one or more computer-readable media as recited in claim 8, wherein the information click conversion rate is obtained by calculating a ratio of a number of times information is exposed to a number of times the information is clicked.

13. The one or more computer-readable media as recited in claim 8, wherein the one or more features further comprise a click feedback rate obtained by calculating a ratio of a number of times feedback is received to a number of times information is clicked.

14. The one or more computer-readable media as recited in claim 8, wherein the one or more features further comprise a timeliness factor indicating a timeliness of the respective inquiry result.

18. The one or more computer-readable media as recited in claim 8, wherein the collecting the active level factor comprises:

obtaining one or more active level indicators of the supplier relating to the respective inquiry result, the one or more active level indicators including one or any combination of;

an online time of the supplier relating to the inquiry result and an active grade of the supplier relating to the inquiry result; and

conducting a linear fitting of the one or more active level indicators to obtain the active level factor.

19. The one or more computer-readable media as recited in claim 8, wherein the collecting the information click conversion rate comprises:

obtaining a number of times the information being exposed and a number of times the information being clicked in a period of time according to statistics of information of a user log; and

calculating a ratio of the number of times the information being exposed to the number of times the information being clicked as the information click conversion rate.

20. The one or more computer-readable media as recited in claim 8, wherein the collecting the feedback conversion rate of information comprises:

obtaining a number of times the information being clicked and a number of times the information with feedbacks in a period of time according to statistics of information of a user log; and

calculating a ratio of the number of times the information with feedbacks to the number of times the information being clicked as the information feedback conversion rate.