MULTI-TIERED APPROACH TO E-MAIL PRIORITIZATION

• US 20130339276A1
• Filed: 06/20/2012
• Published: 12/19/2013
• Est. Priority Date: 02/10/2012
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Lai, Jennifer C., Lu, Jie, Pan, Shimei, Wen, Zhen
• Granted Patent: US 9,256,862 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]

Abstract

A method of automating incoming message prioritization. The method including training a global classifier of a computer system using training data. Dynamically training a user-specific classifier of the computer system based on a plurality of feedback instances. Inferring a topic of the incoming message received by the computer system based on a topic-based user model. Computing a plurality of contextual features of the incoming message. Determining a priority classification strategy for assigning a priority level to the incoming message based on the computed contextual features of the incoming message and a weighted combination of the global classifier and the user specific classifier. Classifying the incoming message based on the priority classification strategy.

Claims

1. A method of automating incoming message prioritization, the method comprising:

training a global classifier of a computer system using training data;

dynamically training a user-specific classifier of the computer system based on a plurality of feedback instances;

inferring a topic of the incoming message received by the computer system based on a topic-based user model;

computing a plurality of contextual features of the incoming message;

determining a priority classification strategy for assigning a priority level to the incoming message based on the computed contextual features of the incoming message and a weighted combination of the global classifier and the user specific classifier; and

classifying the incoming message based on the priority classification strategy.

2. The method according to claim 1, wherein the priority classification strategy is determined from a plurality of priority classification strategies.

3. The method according to claim 2, wherein the plurality of priority classification strategies comprises a dynamic linear combination scheme with instance matching based on instance matching between of the contextual features of the incoming message and contextual features of the plurality of feedback instances, the dynamic liner combination scheme with instance matching comprising:

assessing a quality of the user-specific classifier; and

assigning a weight to each of the global classifier and the user-specific classifier for a linear combination thereof, based on the assessed quality of the user-specific classifier.

4. The method according to claim 2, wherein the plurality of priority classification strategies comprises a dynamic linear combination scheme with instance matching, and

wherein, when the incoming message and a feedback instance of the plurality of feedback instances have at least one of a same sender and subject, the dynamic linear combination scheme with instance matching assigns a same priority to the incoming message as a priority assigned to the feedback instance having at least one of the same sender and subject.

5. The method according to claim 4, wherein, when the incoming message does not have at least one of the same sender and subject as any of the plurality of feedback instances, the dynamic linear combination scheme with instance matching assigns a weight to each of the global classifier and the user-specific classifier for a linear combination thereof.

6. The method according to claim 1, wherein the plurality of contextual features comprises a plurality of people-centric contextual features and a plurality of message-centric contextual features.

7. The method according to claim 1, further comprising updating the user-specific classifier only based on explicit feedback.

8. The method according to claim 1, further comprising updating the user-specific classifier based on explicit feedback and implicit feedback.

9. The method according to claim 1, wherein the plurality of feedback instances includes explicit feedback instances, each explicit feedback instance indicating a priority ranking of a different message having been received by the input.

10. The method according to claim 1, further comprising computing a contextual feature of the plurality of contextual features of the incoming message based on the inferred topic of the incoming message.

11. The method according to claim 1, further comprising:

calculating a first percentage of previously received messages that have a substantially similar topic as the inferred topic of the incoming message;

calculating a second percentage of the previously received messages that have the substantially similar topic which are determined to have been read;

calculating a third percentage of the previously received messages that have the substantially similar topic which are determined to have been at least one of forwarded, replied, saved, and flagged; and

computing a contextual feature of the plurality of contextual features by dynamically combining the first percentage, the second percentage, and the third percentage.

12. The method according to claim 1, further comprising computing the plurality of contextual features of the incoming message based on the inferred topic of the incoming message.

13. The method according to claim 1, further comprising computing a message centric feature of the plurality of contextual features based on a percentage of received messages comprising a substantially similar topic as the inferred topic of the incoming message.

14. A method of automating a prioritization of an incoming message, the method comprising:

training a global classifier of a computer system with training data;

dynamically training a user-specific classifier of a computer system with a plurality of feedback instances;

generating a topic-based user model comp plurality of topic models;

inferring a topic of the incoming message received by the computer system;

computing a plurality of contextual features of the incoming message based on the inferred topic of the incoming message;

dynamically combining the global classifier and the user-specific classifier;

assigning a weight to each contextual feature of the plurality of contextual features based on the dynamic combination of the global classifier and the user specific classifier;

combining the assigned weight of each contextual feature; and

determining a priority level of the incoming message.

15. The method according to claim 14, wherein the determining of the priority level is based on the combined assigned weight of the contextual features.

16. The method according to claim 14, wherein the inferring of the topic of the incoming message is based on one or more topic models of the plurality of topic models.

17. The method according to claim 14, further comprising computing the priority level of the incoming message to be a binary priority level.

18. The method according to claim 14, wherein the computer system comprises a first computer and a second computer, the method further comprising transmitting the priority level having been determined from the first computer to the second computer.

19. The method according to claim 18, further comprising:

receiving the incoming message with the first computer before determining the priority level of the incoming message;

assigning the determined priority level to the incoming message; and

transmitting the incoming message with the priority level assigned thereto to the second computer.

20. A method for automating a prioritization of an incoming message, the method comprising:

training a global classifier of a computer system with training data;

dynamically training a user-specific classifier of the computer system with a plurality of feedback instances;

generating a topic-based user model comprising a plurality of topic models;

inferring a topic of the incoming message received by the computer system;

computing a plurality of contextual features of the incoming message based on the inferred topic of the incoming message;

dynamically combining the global classifier and the user-specific classifier;

assigning a weight to a set of contextual feature of the plurality of contextual features based on the dynamic combination of the global classifier and the user specific classifier;

combining the assigned weight of the set of contextual features; and

determining a priority level of the incoming message.

21. The method according to claim 20, wherein the determining of the priority level is based on the combined assigned weight of the set of contextual features.

22. The method according to claim 20, wherein the inferring of the topic of the incoming message is based on one or more topic models of the plurality of topic models.

23. The method according to claim 20, further comprising computing the priority level of the incoming message to be a binary priority level.

24. A non-transitory tangible computer-readable medium embodying a program of machine-readable instructions executable by a digital processing apparatus to perform an instruction control method of automating a prioritization of an incoming message, the instruction control method comprising:

inferring a topic of the incoming message based on a topic-based user model;

computing a plurality of contextual features of the incoming message based on the inferred topic of the incoming message;

dynamically combining a plurality of classifiers;

assigning a weight to the plurality of contextual features based on the dynamic combination of the plurality of classifiers; and

dynamically determining a priority classification strategy for assigning a priority level to the incoming message based on a combination of assigned weight of the contextual features.

25. The non-transitory tangible computer-readable medium according to claim 24, wherein the plurality of classifiers comprises a user-specific classifier and a global classifier,

wherein the instruction control method further comprises dynamically assessing a quality of the user-specific classifier based on a matching between the incoming message and the plurality of feedback instances, and

wherein, when the matching is above a predetermined threshold, the user-specific classifier is assigned a weight that is greater than a weight assigned to the global classifier.