Sorted Inbox with Important Message Identification Based on Global and User Models

• US 20120054132A1
• Filed: 08/27/2011
• Published: 03/01/2012
• Est. Priority Date: 08/27/2010
• Status: Active Grant

• Current Assignee: Google LLC (Alphabet Inc.)
• Original Assignee: Google Inc. (Alphabet Inc.)
• Inventors: Pietraszek, Tadeusz, Aberdeen, Douglas, de Kloet, David S., Chen, Annie
• Granted Patent: US 8,700,545 B2
• Time in Patent Office: Days
• US Class Current: 706/12
• CPC Class Codes: G06Q 10/10 Office automation; Time management
• CPC Class Codes: G06Q 10/107 Computer-aided management of electronic mailing [e-mailing]

Abstract

A server system receives a message associated with a user. The server extracts message signals from the message, the extracted message signals including a first plurality of message signals. The server generates an importance weight for each message signal of the first plurality of message signals using both a global importance prediction model and a user importance prediction model. The global importance prediction model is based on information associated with multiple users, and the user importance prediction model is based on information associated with the user. The server determines an importance score for the message based on the generated importance weights of the first plurality of message signals. The server sends message information for the message along with information regarding importance of the message for display at a client device, where the information regarding importance of the message is based at least on the determined importance score.

Claims

1. A method, comprising:

at a server system having one or more processors and memory storing one or more programs in the memory to be executed by the one or more processors, the one or more programs including instructions for;

receiving a message associated with a user;

extracting message signals from the message, the extracted message signals including a first plurality of message signals;

generating an importance weight for each message signal of the first plurality of message signals, the generating including, for a respective message signal;

determining a first weight for the respective message signal using a first importance prediction model;

determining a second weight for the respective message signal using a second importance prediction model;

determining the importance weight of the respective message signal by combining the first weight and the second weight;

wherein the first importance prediction model is based on information associated with multiple users, and the second importance prediction model is based on information associated with the user; and

determining an importance score for the message based on the generated importance weights of the first plurality of message signals;

sending the message along with information regarding importance of the message to the user for display at a client device, wherein the information regarding importance of the message is based at least on the determined importance score.

2. The method of claim 1,

wherein the extracted message signals comprises the first plurality of message signals and a second plurality of message signals;

the method further comprising;

generating an importance weight for each message signal of the second plurality of message signals using the second importance prediction model; and

determining the importance score for the message based on the generated importance weights of the first plurality of message signals and the generated importance weights of the second plurality of message signals.

3. The method of claim 1, further including comparing the importance score of the message with a threshold to determine importance of the message.

4. The method of claim 3, further including periodically updating the threshold using machine learning.

5. The method of claim 1, further comprising collecting feedback data from the user regarding importance of the message, and modifying the second importance prediction model using the feedback data.

6. The method of claim 1, wherein combining the first weight and the second weight comprises adding the first weight and the second weight.

7. The method of claim 1, wherein the first importance prediction model and second importance prediction model both include a plurality of term-related weights, each for a distinct term-related message signal corresponding to presence or quantity of important terms in the message, wherein important terms comprise terms determined to be indicative of message importance.

8. The method of claim 1, wherein

the message includes information identifying a set of message participants,

the user has an associated social graph having a set of social graph members;

and

the first importance prediction model and second importance prediction model both include a plurality of social graph-related weights, each concerning at least one of;

presence of social graph members in the message participants, interactions of one or more social graph members with the message, interactions of one or more social graph members with information having a predefined relationship to the message.

9. The method of claim 1, further including updating one or more of the importance weights in the first importance prediction model and second importance prediction model using a time-dependent decay function.

10. The method of claim 1, further including automatically generating the importance weights for both the first importance prediction model and the second importance prediction model without collecting feedback data from the user regarding importance of any messages.

11. The method of claim 1, further including automatically generating the importance weights for both the first importance prediction model and the second importance prediction model without requiring the user to provide feedback data regarding importance of any messages.

12. The method of claim 1, further including automatically generating the importance weights for the first importance prediction model while excluding user provided message importance feedback, if any;

and automatically generating the importance weights for the second importance prediction model while incorporating optional user provided message importance feedback, if any.

13. The method of claim 1, further including periodically updating one or more of the importance weights in the first and second importance prediction model using machine learning.

14. The method of claim 1, wherein the first plurality of message signals include one or more combination message signals, wherein the combination message signals are combination of two or more message signals.

15. A server system, comprising:

one or more processors;

memory storing one or more programs for execution by the one or more processors, the one or more programs including instructions for;

receiving a message associated with a user;

extracting message signals from the message, the extracted message signals including a first plurality of message signals;

generating an importance weight for each message signal of the first plurality of message signals, the generating including, for a respective message signal;

determining a first weight for the respective message signal using a first importance prediction model;

determining a second weight for the respective message signal using a second importance prediction model;

determining the importance weight of the respective message signal by combining the first weight and the second weight;

wherein the first importance prediction model is based on information associated with multiple users, and the second importance prediction model is based on information associated with the user; and

determining an importance score for the message based on the generated importance weights of the first plurality of message signals;

sending message information for the message, comprising the message or information identifying the message, along with information regarding importance of the message to the user for display at a client device, wherein the information regarding importance of the message is based at least on the determined importance score.

16. The server system of claim 15,

wherein the extracted message signals comprises the first plurality of message signals and a second plurality of message signals;

the one or more programs further comprising instructions for;

generating an importance weight for each message signal of the second plurality of message signals using the second importance prediction model; and

determining the importance score for the message based on the generated importance weights of the first plurality of message signals and the generated importance weights of the second plurality of message signals.

17. The server system of claim 15, the one or more programs further comprising instructions for comparing the importance score of the message with a threshold to determine importance of the message.

18. The server system of claim 15, wherein

the message includes information identifying a set of message participants,

the user has an associated social graph having a set of social graph members;

and

the first importance prediction model and second importance prediction model both include a plurality of social graph-related weights, each concerning at least one of;

presence of social graph members in the message participants, interactions of one or more social graph members with the message, interactions of one or more social graph members with information having a predefined relationship to the message.

19. A non-transitory computer readable storage medium storing one or more programs for execution by the one or more processors of a server system, the one or more programs including instructions for:

receiving a message associated with a user;

extracting message signals from the message, the extracted message signals including a first plurality of message signals;

generating an importance weight for each message signal of the first plurality of message signals, the generating including, for a respective message signal;

determining a first weight for the respective message signal using a first importance prediction model;

determining a second weight for the respective message signal using a second importance prediction model;

determining the importance weight of the respective message signal by combining the first weight and the second weight;

wherein the first importance prediction model is based on information associated with multiple users, and the second importance prediction model is based on information associated with the user; and

determining an importance score for the message based on the generated importance weights of the first plurality of message signals;

sending the message along with information regarding importance of the message to the user for display at a client device, wherein the information regarding importance of the message is based at least on the determined importance score.

20. The non-transitory computer readable storage medium of claim 19,

wherein the extracted message signals comprises the first plurality of message signals and a second plurality of message signals;

the one or more programs further comprising instructions for;

generating an importance weight for each message signal of the second plurality of message signals using the second importance prediction model; and

determining the importance score for the message based on the generated importance weights of the first plurality of message signals and the generated importance weights of the second plurality of message signals.

21. The non-transitory computer readable storage medium of claim 19, the one or more programs further comprising instructions for comparing the importance score of the message with a threshold to determine importance of the message.

22. The non-transitory computer readable storage medium of claim 19, wherein

the message includes information identifying a set of message participants,

the user has an associated social graph having a set of social graph members;

and

the first importance prediction model and second importance prediction model both include a plurality of social graph-related weights, each concerning at least one of;

presence of social graph members in the message participants, interactions of one or more social graph members with the message, interactions of one or more social graph members with information having a predefined relationship to the message.