MINING EMAIL INBOXES FOR SUGGESTING ACTIONS

• US 20120290662A1
• Filed: 05/11/2011
• Published: 11/15/2012
• Est. Priority Date: 05/11/2011
• Status: Active Grant

• Current Assignee: Verizon Media, Inc. (Verizon Communications Inc.), Yahoo Assets LLC
• Original Assignee: Yahoo! Inc. (Apollo Global Management, Inc.)
• Inventors: Maarek, Yoelle, Weber, Ingmar, Koren, Yehuda Arie
• Granted Patent: US 10,366,341 B2
• Time in Patent Office: Days
• US Class Current: 709/206
• CPC Class Codes: G06F 16/2465 Query processing support for facilitating data mining operations in structured databases
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]

Abstract

Embodiments are directed towards automatically learning user behavioral patterns when interacting with messages and based on the learned patterns, suggesting one or more predicted actions that a user might take in response to receiving subsequent message. One or more classifiers are trained and employed to predict one or more actions that a user might take in response to receiving the message. In one embodiment, the one or more predicted actions are provided suggested to the user as an action the user might take on the message. Messages may be rank ordered within a given suggested action based on a confidence level of the prediction.

Claims

1. A network device, comprising:

a transceiver to send and receive messages over a network; and

a processor that is operative on the received messages to perform actions, including;

receiving a message directed towards a user;

employing at least one binary machine learning classifier to automatically predict at least one action from a plurality of actions that the user might be expected to take on the received message; and

displaying to the user the message with the predicted action as a suggested action for the user to take on the message.

2. The network device of claim 1, wherein employing the at least one binary machine learning classifier further comprises:

training at least three binary machine learning classifiers, wherein the classifiers comprise at least one horizontal classifier, a vertical general classifier, and a vertical pair-wise classifier; and

selectively employing the at least three classifiers, based on whether the user is determined to be a new user to the classifiers or has provided historical actions to previously received messages, the historical actions being available to train at least one of the vertical classifiers.

3. The network device of claim 1, wherein at least two suggested actions are predicted and displayed for the message.

4. The network device of claim 1, wherein at least one of the actions in the plurality of actions is directed towards preserving a conversation associated with the message.

5. The network device of claim 1, wherein the processor is operative to perform actions, further including:

determining a confidence value in the predicted action for the message; and

employing the confidence value to generate and display the message within a rank ordering of a plurality of other messages having a same suggested action.

6. The network device of claim 1, wherein at least one binary machine learning classifier is configured to be retrained based on an action actually taken by the user on the message.

7. The network device of claim 1, wherein displaying of the suggested action further comprises displaying a selectable icon, such that when the selectable icon is selected, the suggested action is initiated for at least the message.

8. A method operating on one or more network devices, the method comprising:

receiving a plurality of messages directed towards a user, at least one message in the plurality being received from a sender that is different from an other sender associated with an other message in the plurality of messages;

selecting one or more classifiers based in part based on whether the user is determined to be a new user, and whether the user has received messages previously from one of the sender or the other sender;

employing the selected one or more classifiers to automatically predict at least one action from a plurality of actions that the user is likely to take on each of the of the plurality of messages; and

displaying to the user each of the plurality of messages with the predicted action for each message where the predicted action is identified as a suggested action to be taken by the user.

9. The method of claim 8, wherein the one or more classifiers comprise at least a horizontal classifier, a vertical general classifier, and a vertical pair-wise classifier.

10. The method of claim 8, wherein selecting one or more classifiers further comprises:

determining whether the user is the new user, and if so, then employing a horizontal classifier that is trained to predict how a plurality of users might respond to a given message;

determining whether the user has received messages from the sender or the other sender previously, and if so, then employing a vertical pair-wise classifier that is trained to predict how this user is likely to respond to a given message from the sender or the other sender; and

determining that the user is not a new user, and has not received messages from the sender or the other sender previously, and employing based on the determination, a vertical general classifier that is trained to predict how this user is likely to response to a given message generally.

11. The method of claim 8, wherein at least two suggested actions are displayed for at least one message.

12. The method of claim 8, wherein messages having a same suggested action are rank ordered within the same suggested action based on a confidence value associated with a confidence of a prediction in the suggested action.

13. The method of claim 8, further comprising:

predicting how the user is likely to respond to a particular message having an advertisement, and based on the prediction, selectively sending the particular message to the user.

14. The method of claim 8, wherein employing the selected one or more classifiers further comprises combining weighted results from the selected one or more classifiers to predict the one or more actions.

15. A computer-readable storage device having computer-executable instructions stored thereon such that when the storage device is accessed by a computing device, the instructions are executable by the computing device to perform actions, comprising:

receiving a plurality of messages directed towards a user, at least one message in the plurality being received from a sender that is different from an other sender associated with an other message in the plurality of messages;

selecting one or more classifiers based in part based on whether the user is determined to be a new user, and whether the user has received messages previously from one of the sender or the other sender;

employing the selected one or more classifiers to automatically predict at least one action from a plurality of actions that the user is likely to take on each of the of the plurality of messages; and

displaying to the user each of the plurality of messages with the predicted action for each message where the predicted action is identified as a suggested action to be taken by the user.

16. The computer-readable storage device of claim 15, further comprising:

predicting how the user is likely to respond to a particular message having an advertisement, and based on the prediction, selectively sending the particular message to the user.

17. The computer-readable storage device of claim 15, wherein messages having a same suggested action are rank ordered within the same suggested action based on a confidence value associated with a confidence of a prediction in the suggested action.

18. The computer-readable storage device of claim 15, wherein at least two suggested actions are displayed for at least one message.

19. The computer-readable storage device of claim 15, wherein selecting one or more classifiers further comprises:

determining whether the user is the new user, and if so, then employing a horizontal classifier that is trained to predict how a plurality of users might respond to a given message;

determining whether the user has received messages from the sender or the other sender previously, and if so, then employing a vertical pair-wise classifier that is trained to predict how this user is likely to respond to a given message from the sender or the other sender; and

determining that the user is not a new user, and has not received messages from the sender or the other sender previously, and employing based on the determination, a vertical general classifier that is trained to predict how this user is likely to response to a given message generally.

20. computer-readable storage device of claim 15, wherein the one or more classifiers comprise at least a horizontal classifier, a vertical general classifier, and a vertical pair-wise classifier.