SCALABLE MODEL-BASED PRODUCT MATCHING

• US 20080313165A1
• Filed: 06/15/2007
• Published: 12/18/2008
• Est. Priority Date: 06/15/2007
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Wu, Kai, Yang, Hua, Wen, Ji-Rong, Reel, Eric Anthony, Yao, Tong, Zhang, Jiyu, Takacs, Daniel, Hart, Jonathan R.M.
• Granted Patent: US 7,979,459 B2
• Time in Patent Office: Days
• US Class Current: 707/5
• CPC Class Codes: G06Q 10/101 Collaborative creation, e.g. joint development of products or services
• CPC Class Codes: G06Q 30/02 Marketing; Price estimation or determination; Fundraising

Abstract

Aspects of the subject matter described herein relate to matching product information to products. In aspects, a product matching component receives product information. The product matching component normalizes the product information and obtains keywords from the product information. By querying a database of recognized products, the keywords are used to obtain a list of products that potentially match the product information. A confidence level is assigned to each of the potential matches in the list. A match may be returned for the highest matched product or for a selectable number of products whose confidence level(s) exceed a selectable threshold.

Claims

1. A computer-readable medium having computer-executable instructions, which when executed perform actions, comprising:

receiving product information that includes data regarding a product;

obtaining a list of one or more products that potentially match the product information;

assigning a confidence level to at least one of the one or more products, at least one of the products having a highest of the confidence levels;

determining whether any of the products match the product information based on the confidence level;

if at least one of the products match the product information, indicating that a match was found; and

if all of the products do not match the product information, indicating that no match was found.

2. The computer-readable medium of claim 1, further comprising identifying key information within the product information and wherein obtaining a list of one or more products that potentially match the product information comprises using the key information to obtain the list from a database of recognized products.

3. The computer-readable medium of claim 2, wherein identifying key information within the product information comprises identifying words within the product information that are more indicative of a product based on a frequency of the words within a model that includes pre-computed data regarding the frequency of the words.

4. The computer-readable medium of claim 1, further comprising ranking the one or more products based on their corresponding confidence levels.

5. The computer-readable medium of claim 1, wherein the product information comprises fields some of which potentially do not include data.

6. The computer-readable medium of claim 5, further comprising:

for each non-empty field in the product information;

calculating a distance between the field in the product information and a corresponding field of the product,

multiplying the field by a coefficient to obtain a value, and

adding the value to a running total; and

calculating the confidence level based on the running total.

7. The computer-readable medium of claim 1, further comprising normalizing tokens within the product information.

8. The computer-readable medium of claim 7, wherein normalizing tokens within the product information comprises removing characters indicated by a model.

9. The computer-readable medium of claim 7, wherein normalizing tokens within the product information comprises replacing a first phrase within the product information with a second phrase.

10. The computer-readable medium of claim 1, wherein determining whether any of the products match the product information based on the confidence level comprises determining whether at least one of the products has a confidence level greater than a threshold, the threshold being selectable.

11. The computer-readable medium of claim 1, wherein indicating that a match was found comprises providing an identifier of a product having the highest confidence level

12. A method implemented at least in part by a computer, the method comprising:

receiving product information that includes data regarding a product;

identifying at least one keyword within the product information based on a model that indicates a correlation between the at least one keyword and a product;

providing the at least one keyword to a database that includes recognized products; and

receiving a list of products that includes the at least one keyword.

13. The method of claim 12, wherein the model is generated via data from at least one category of products.

14. The method of claim 12, wherein the model includes an inverse frequency of keywords associated with products.

15. The method of claim 12, further comprising removing data from the product information that is deemed less significant in matching the product information to a product.

16. The method of claim 15, wherein removing data from the product information that is deemed less significant in matching the product information to a product comprises removing the data based on a model that utilizes bigrams or n-grams to remove less relevant words.

17. The method of claim 12, further comprising selecting one or more keywords from the at least one keyword based on values associated with the one or more keywords and a threshold, the selecting one or more keywords comprising selecting additional keywords from the at least one keyword until a sum of values of the selected keywords exceeds a threshold.

18. In a computing environment, an apparatus, comprising:

a store capable of storing a model that is related to one or more product categories;

a normalizer operable to receive a first set of product information and to provide a second set of product information product information, the second set of product information being normalized per the model; and

a key term selector operable to extract and select keywords from the product information, the keywords usable to lookup potential matches from a product database.

19. The apparatus of claim 18, wherein the model comprises a list of phrases that are to be replaced by other phrases by the normalizer.

20. The apparatus of claim 18, wherein the model comprises a table of inverted word frequencies usable to eliminate less significant data in the product information.