Methods for routing items for communications based on a measure of criticality

• US 7,337,181 B2
• Filed: 07/15/2003
• Issued: 02/26/2008
• Est. Priority Date: 07/30/1999
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric
• Primary Examiner(s): WOO, ISAAC M
• Application Number:
  US10/620,116
• Publication Number:
  US 20040015557A1
• Time in Patent Office: 1,687 Days
• Field of Search: 707 1- 10, 707100-1041, 707200-206, 709204-207, 709238-244
• US Class Current: 1/1
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]
• CPC Class Codes: H04L 51/214 using selective forwarding
• CPC Class Codes: H04L 51/224 providing notification on incoming messages, e.g. pushed notifications of received messages
• CPC Class Codes: H04L 51/226 Delivery according to priorities
• CPC Class Codes: Y10S 707/99936 Pattern matching access
• CPC Class Codes: Y10S 707/99939 Privileged access
• CPC Class Codes: Y10S 707/99942 Manipulating data structure, e.g. compression, compaction, compilation
• CPC Class Codes: Y10S 707/99943 Generating database or data structure, e.g. via user interface
• CPC Class Codes: Y10S 707/99944 Object-oriented database structure
• CPC Class Codes: Y10S 707/99945 Object-oriented database structure processing

Abstract

The routing of prioritized documents such as email messages is disclosed. In one embodiment, a computer-implemented method first receives a text. The method assigns a priority to the document, based on a text classifier such as a Bayesian classifier or a support-vector machine classifier. The method then routes the text based on a routing criteria. In one embodiment the routing is directed by a measure of priority that reflects the expected cost of delayed review of the document.

Claims

1. A computer-implemented method utilizing a probabilistic-based classifier trained with predefined data sets that are indicative of item priority levels, comprising:

implicitly training the probabilistic-based classifier to infer a priority level of a received item based in part on at least one of current or historical information of at least a focus of attention of a user that are indicative of item priority levels, the focus of attention comprising at least one of keyboard activity or mouse activity, or a combination thereof, associated with the user;

determining a priority level of the received item utilizing the probabilistic-based classifier, the priority being representative of at least an urgency of the received item relative to the intended recipient, the priority comprises a measure of a rate of cost accrued with delayed review of the received item; and

utilizing the priority level to facilitate electronic communication.

2. The method of claim 1, the probabilistic-based classifier is at least one of a Bayesian classifier and a support-vector machine classifier.

3. The method of claim 1, the probabilistic-based classifier is explicitly trained.

4. The method of claim 3, the explicit training is performed during initial phases of constructing the probabilistic-based classifier.

5. The method of claim 3, the predefined data set employed for explicitly training the probabilistic-based classifier comprises a training set to discriminate between time-critical and non-time-critical items.

6. The method of claim 3, explicitly training the probabilistic-based classifier comprises utilizing feature selection.

7. The method of claim 6, the feature selection includes a mutual information analysis.

8. The method of claim 6, the feature selection operates on single words.

9. The method of claim 6, the feature selection operates on phrases.

10. The method of claim 6, the feature selection operates on parts of speech.

11. The method of claim 6, the feature selection employs high-level patterns.

12. The method of claim 6, the feature selection utilizes tokens.

13. The method of claim 6, the feature selection utilizes tagged text to discriminate features of the received item.

14. The method of claim 1, the probabilistic-based classifier is implicitly trained.

15. The method of claim 14, further comprising implicitly training the probabilistic-based classifier based at least in part on an assumption that time-critical items are reviewed prior to non-time-critical items.

16. The method of claim 14, farther comprising continually updating the probabilistic-based classifier via the implicit training.

17. The method of claim 1, further comprising implicitly training the probabilistic-based classifier based on at least one of current or historical information of at least one of user presence or a focus of attention of a user.

18. A computer-implemented method, comprising:

determining a loss function based on an expected cost in lost opportunities as a function of an amount of time delayed in reviewing an item after the item has been received, the lost opportunities comprising an opportunity to attend a meeting at a specified time;

classifying priority of the item based in part on the loss function utilizing a trained classifier; and

utilizing the classified priority of the item to infer a desired computer-based automated action to take to facilitate electronic communication.

19. The method of claim 18, the trained classifier is at least one of a Bayesian classifier or a support-vector machine classifier, or a combination thereof.

20. The method of claim 18, the trained classifier classifies the priority of the item based on a loss function.

21. The method of claim 20, the loss function is determined based on an expected cost in lost opportunities as a function of an amount of time delayed in reviewing the item after it has been received.

22. The method of claim 20, the loss function is determined based on a type of the item.

23. The method of claim 20, the loss function is at least one of a linear loss function or a non-linear loss function, or a combination thereof.

24. The method of claim 18, the trained classifier is explicitly trained.

25. The method of claim 24, the explicit training is performed during the construction of the trained classifier.

26. The method of claim 24, the explicit training employs a predefined training set of data to discriminate between time-critical and non-time-critical items.

27. The method of claim 24, explicitly training the trained classifier comprises utilizing feature selection.

28. The method of claim 27, the feature selection operates on at least one of single words, phrases, or parts of speech, or a combination thereof.

29. The method of claim 27, the feature selection utilizes at least one of tokens or tagged text, or a combination thereof.

30. The method of claim 18, the trained classifier is implicitly trained.

31. The method of claim 30, the trained classifier is implicitly trained based on at least one of current or historical information of at least one of user presence, activity of a user, or a focus and attention of the user.

32. The method of claim 30, the trained classifier is implicitly trained based at least in part on an assumption that time-critical items are reviewed prior to non-time-critical items.

33. The method of claim 30, the trained classifier is continually updated via the implicit training.