Machine learning system and method comprising segregator convergence and recognition components to determine the existence of possible tagging data trends and identify that predetermined convergence criteria have been met or establish criteria for taxonomy purpose then recognize items based on an aggregate of user tagging behavior

• US 7,672,909 B2
• Filed: 12/20/2006
• Issued: 03/02/2010
• Est. Priority Date: 09/28/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Cheng, Lili, MacLaurin, Matthew B., Meijer, Henricus Johannes Maria, Ozzie, Raymond E., Flake, Gary W., Zaner-Godsey, Melora, Brumme, Christopher W., Blinn, Arnold N., Gounares, Alexander G., Glasser, Daniel S., Bergstraesser, Thomas F., Mishra, Debi P., Connolly, Michael
• Primary Examiner(s): Vincent; David R
• Assistant Examiner(s): Brown, Jr.; Nathan H
• Application Number:
  US11/613,779
• Publication Number:
  US 20080082466A1
• Time in Patent Office: 1,168 Days
• Field of Search: 706/12
• US Class Current: 706/12
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 20/10 using kernel methods, e.g. support vector machines [SVM]
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks

Abstract

Systems and methods that recognize items based on aggregated tagging behavior of users regarding the items. Individual users can establish relationships among tagged items and the system analyzes aggregate of such established relationships, to recognize the item and/or infer additional information regarding the items (e.g., wisdom of crowd such as metadata annotations, relevance ranking, and the like). Subsequently, search engine crawlers can be supplied with such additional information (e.g., extra metadata for reverse link search tables) to facilitate enterprise management and search.

Claims

1. A machine learning system that analyzes tagging behavior, comprising:

a segregator component configured to initially perform a determination of an existence of a possible trend of tagging data based on collective user tagging behavior among a plurality of items, the determination including an extraction of one or more tags derived from analysis of the plurality of items to establish the possible trend;

a convergence component configured to analyze the possible trend to identify that a predetermined convergence criteria has been met or to establish a criteria for taxonomy purpose based, at least in part, upon tagged items and user tagging behavior relationships between the plurality of items;

a recognition component configured to recognize the plurality of items based on aggregate of user tagging behavior associated with the items; and

a storage medium that stores recognized items.

2. The machine learning system of claim 1, the recognition component further comprising generic classifiers that classify tagged items based on aggregate of user tagging behavior.

3. The machine learning system of claim 1, the recognition component further comprising user specific classifier that is based on user demographics.

4. The machine learning system of claim 2 further comprising a search engine that searches for items based on classification.

5. The machine learning system of claim 4 further comprising a Dewey Decimal System that facilitates data organization based on hierarchical tags.

6. The machine learning system of claim 1 further comprising a training model for determination of tagging trends.

7. The machine learning system of claim 3 further comprising an artificial intelligence component that facilitates trend establishment among tagged items, the tagged items including at least one of digital photographs, e-mails, web pages, and data files.

8. The machine learning system of claim 1, further comprising a display component having a tag controller that sets thresholds for trend establishment.

9. The machine learning system of claim 4 further comprising a Bayesian classifier.

10. The machine learning system of claim 1 further comprising a user feed back component that collects user feedback associated with tagging behavior of items.

11. A method of inferring tagging behavior comprising:

collecting, by a machine learning system configured to recognize items via tagging behavior, relationships established by a plurality of users when tagging items;

displaying, by the machine learning system, an interface to a user to select one or more tags and to adjust how closely the one or more tags relate to a relationship; and

recognizing, by the machine learning system, the items based on tagging behavior of users.

12. The method of claim 11 further comprising evaluating, by the machine learning system, a trustworthiness of a user based on relationships established thereby.

13. The method of claim 11 further comprising inferring, by the machine learning system, relationships among tags.

14. The method of claim 11 further comprising organizing, by the machine learning system, items based on tagging behavior of the plurality of users.

15. The method of claim 11 further comprising browsing, by the machine learning system, data based on algorithmically-deduced relationships between items.

16. The method of claim 11 further comprising searching, by the machine learning system, for items based on the tagging behavior of users.

17. The method of claim 11 further comprising training, by the machine learning system, a machine learning system for inferring tagging behavior of users.

18. The method of claim 11 further comprising automatically generating, by the machine learning system, possible tagging criteria.

19. The method of claim 11 further comprising incorporating, by the machine learning system, inferred tagging criteria in taxonomy applications.

20. A computer storage medium encoded with processor-executable instructions that, when executed, direct a computing system to perform a method, the method comprising:

collecting information regarding tagged items from a plurality of users;

determining an existence of one or more trends among tags and associated items based on a predetermined criteria and an inferred relationship between the tagged items in an event that the predetermined criteria does not exist;

examining convergence among the plurality of trends to identify that a predetermined convergence criteria has been met or to establish a new convergence criteria for taxonomy purposes in an event that the predetermined convergence criteria does not exist; and

identifying a tagging criteria from the examination for taxonomy purposes.