Make and model classifier

• US 8,392,432 B2
• Filed: 04/12/2010
• Issued: 03/05/2013
• Est. Priority Date: 04/12/2010
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Li, Xiaolei, Wu, Lingfeng, Dai, Wenjun, Wu, Xiao, Ha, Vu Anh, Zhu, Weibin
• Primary Examiner(s): Perveen, Rehana
• Assistant Examiner(s): BUI, THUY T
• Application Number:
  US12/758,539
• Publication Number:
  US 20110252042A1
• Time in Patent Office: 1,058 Days
• Field of Search: 707/748, 707/771, 707/777, 707/778
• US Class Current: 707/748
• CPC Class Codes: G06F 16/313 Selection or weighting of terms for indexing

Abstract

Systems and methods for assigning a product category and/or classification to user data based on a taxonomy of makes and models is provided. User data is received and converted to plain text. A taxonomy is provided, having a plurality of categories corresponding with product descriptions based on product makes and models. Rules are associated with the categories in the taxonomy and used to generate a category score for categories in the taxonomy. The category scores are used to assign at least one category from the taxonomy to the user data. A product class may also be assigned to the user data based on the assigned product category.

Claims

1. One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, causes the one or more computing devices to perform a method comprising:

providing a taxonomy having a plurality of categories, each category corresponding with a product description based on product makes and models;

associating one or more rules with each category in the taxonomy;

receiving user data;

converting the user data into plain text;

generating a category score for each of one or more categories in the taxonomy by comparing the plain text to one or more rules of each of the one or more categories; and

assigning at least one category from the one or more categories to the user data based on the category score for each of the one or more categories;

wherein generating a category score for each of the one or more categories comprises;

determining a category score for each category in a current level in the taxonomy;

selecting one or more top confident categories from the current level by comparing the category scores generated for each category in the current level in the taxonomy with one or more thresholds;

determining that at least one top confident category includes at least one child category at a next level in the taxonomy;

generating a category score for each child category associated with the at least one top confident category by comparing the plain text to one or more rules for each child category;

selecting one or more top confident categories from the at least one child category; and

repeating determining a category score for each category in a current level in the taxonomy, selecting one or more top confident categories from the current level, determining that at least one top confident category includes at least one child category at a next level in the taxonomy, generating a category score for each child category associated with the at least one top confident category, and selecting one or more top confident categories from the at least one child category, until it is determined that no top confident categories include at least one child category.

2. The one or more computer storage media of claim 1, wherein each rule associated with each category in the taxonomy comprises a pair of a term associated with a weight.

3. The one or more computer storage media of claim 2, wherein associating one or more rules with each category in the taxonomy includes sorting the one or more rules by a number of terms in the one or more rules.

4. The one or more computer storage media of claim 3, wherein comparing the plain text to one or more rules of the one or more categories includes determining if one or more terms in the plain text match one or more terms in the rules.

5. The one or more computer storage media of claim 1, wherein the user data is received from one or more selected from the following:

a search query; and

a webpage view.

6. The one or more computer storage media of claim 1, wherein the method further comprises:

based on assigning the at least one category to the user data, assigning one or more classes to the user data.

7. The one or more computer storage media of claim 1, wherein the method further comprises:

generating a result list that includes the category scores for each of the one or more categories.

8. The one or more computer storage media of claim 7, wherein the result list is sorted in descending order.

9. A computer system executed by one or more computer processors, comprising:

one or more computing devices, wherein each of the one or more computing devices has a processor and a memory, and further wherein the one or more computing devices comprises a user data component, an assigning component, and a reporting component,

wherein the user data component is configured to receive user data and convert the user data into plain text,

the assigning component is configured to assign at least one category from a plurality of categories in a taxonomy to the user data based on the plain text, each category corresponding with a product description based on product makes and models,

and the reporting component is configured to report the at least one category assigned to the user data,

wherein the assigning component assigns at least one category from the plurality of categories in a taxonomy by;

associating one or more rules with each category in the taxonomy;

generating a category score for each of one or more categories in the taxonomy by comparing the plain text to one or more rules of the one or more categories; and

assigning at least one category from the one or more categories to the user data based on the category score for each of the one or more categories,

and further wherein generating a category score for each of one or more categories comprises;

determining a category score for each category in a current level in the taxonomy;

selecting one or more top confident categories from the current level by comparing the category scores generated for each category in the current level in the taxonomy with one or more thresholds;

determining that at least one top confident category includes at least one child category at a next level in the taxonomy;

generating a category score for each child category associated with the at least one top confident category by comparing the plain text to one or more rules for each child category;

selecting one or more top confident categories from the at least one child category; and

repeating determining a category score for each category in a current level in the taxonomy, selecting one or more top confident categories from the current level, determining that at least one top confident category includes at least one child category at a next level in the taxonomy, generating a category score for each child category associated with the at least one top confident category, and selecting one or more top confident categories from the at least one child category until it is determined that no top confident categories include at least one child category.

10. The system of claim 9, wherein the one or more rules of the one or more categories in the taxonomy comprise one or more pairs of a term associated with a weight.

11. The system of claim 10, wherein the one or more rules of one or more categories are sorted by a number of words in the terms.

12. The system of claim 9, wherein the reporting component generates a result list that includes category scores for each of one or more categories.

13. The system of claim 9, wherein the user data is received from one or more selected from the following:

a search query and a web page view.

14. The system of claim 9, wherein the assigning component assigns at least one class to the user data.

15. The system of claim 14, wherein the reporting component reports the at least one class assigned to the at least one category assigned to the user data.

16. One or more computer storage media storing computer-useable instructions that, when used by one or more computing devices, cause the one or more computing devices to perform a method comprising:

providing a taxonomy having a plurality of categories, each category corresponding with a product description based on product makes and models;

associating one or more rules with each category in the taxonomy, each rule comprising one or more pairs of a term associated with a weight;

receiving user data, wherein the user data is received from one or more selected from the following;

a search query; and

a web page view;

converting the user data to plain text;

sorting the one or more rules for each category in the taxonomy by a number of words in one or more terms;

generating a category score for each category in a current level in a taxonomy by comparing the plain text to one or more rules of each of the one or more categories;

assigning at least one category from the one or more categories to the user data based on the category score for each category in the current level;

selecting one or more top confident categories for the current level by comparing the category scores for each category in the current level in the taxonomy with one or more thresholds;

determining that at least one top confident category includes at least one child category at a next level in the taxonomy;

generating a category score for each child category associated with the at least one top confident category by comparing the plain text to one or more rules for each child category;

assigning at least one child category from the current level to the user data based on the category score for the at least one child category;

selecting one or more top confident categories from the at least one child category;

assigning at least one class to the user data based on the at least one assigned category and the at least one assigned child category; and

repeating generating a category score for each category in a current level in the taxonomy, assigning at least one category from the one or more categories to the user data, selecting one or more top confident categories from the current level, determining that at least one top confident category includes at least one child category at a next level in the taxonomy, generating a category score for each child category associated with the at least one top confident category, selecting one or more top confident categories from the at least one child category, and assigning at least one class to the user data based on the at least one assigned category and the at least one assigned child category until it is determined that no top confident categories include at least one child category.

17. The one or more computer storage media of claim 16, wherein comparing the plain text to one or more rules of the one or more categories includes determining if one or more terms in the plain text match one or more terms in the rules.