Principals and methods for balancing the timeliness of communications and information delivery with the expected cost of interruption via deferral policies

• US 7,529,683 B2
• Filed: 06/29/2005
• Issued: 05/05/2009
• Est. Priority Date: 06/29/2005
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J., Achlioptas, Dimitris
• Primary Examiner(s): Hayes; John W
• Assistant Examiner(s): Erb; Nathan
• Application Number:
  US11/171,868
• Publication Number:
  US 20070004385A1
• Time in Patent Office: 1,406 Days
• Field of Search: 705/1
• US Class Current: 705/1.1
• CPC Class Codes: H04L 51/214 using selective forwarding
• CPC Class Codes: H04L 51/226 Delivery according to priorities
• CPC Class Codes: H04L 51/234 for tracking messages

Abstract

A system is provided that facilitates information processing for a user. The system includes an analyzer to automatically determine one or more states of a user'"'"'s availability based on patterns of the user'"'"'s interruptability. A deferral component generates or forwards messages or alerts to the user in accordance with a cost-minimization analysis and determined availability. For example, the deferral component can process interrupts to the user and determine at least one of optimal and approximately optimal deferral times in which to direct an information item to the user, the information item being one of an email message, an incoming phone call, a push-to-talk message, an instant message, an alert, requests, offers of assistance, status or error reports, or a task from another application or autonomous system.

Claims

1. A computer-implemented system that facilitates information processing, comprising:

a processor; and

a memory, communicatively coupled to the processor, having stored thereon at least one component comprising;

an analyzer that automatically determines a user'"'"'s availability based at least in part upon patterns of the user'"'"'s interruptability; and

a deferral component that determines a bounded deferral period at the end of which to deliver an information item to the user in accordance with the user'"'"'s availability, wherein a maximal deferral time is determined based at least in part on one or more of an exact or approximate minimization of a net expected cost to the user both of delay and of interruption, the net expected cost of delay and the net expected cost of interruption scaled to a substantially similar scale, the maximal deferral time, t, being determined via one of an ideal or approximate minimization of the following formula;

W(t)=(c+ƒ

(t))g(t)+∫

₀^t−

g′

(s)ƒ

(s)ds,

where W(t) is a total expected cost;

g(t) describes a probability that the user will be busy after waiting up until the maximal deferral time t;

ƒ

(t) is a loss in the value of seeing the information item, relative to an initial value of reviewing the information item when it arrives, when reviewing the information item at the maximal deferral time t;

ƒ

(s) is a loss of value of the information item at time s that the user might become available before the maximal deferral time t under uncertainty; and

c is a cost of interruption if the user is busy.

2. The system of claim 1, further comprising a log that records interruption patterns of the user over time, the log recording data from local and/or remote network locations, and the interruption patterns including patterns of availability, unavailability, busyness, application activity, user activity, subject matter being worked on, type or numbers of tasks processed, times of busyness, and times when email or voice mail are processed.

3. The system of claim 2, wherein the analyzer processes data from a database via one or more models that are employed to determine and predict user busyness states including when the cost to interrupt the user would be lower versus periods of higher cost when the user is focused on other tasks.

4. The system of claim 1, wherein the deferral component processes interrupts to the user and determines at least one of optimal and approximately optimal deferral times in which to direct the information item to the user, the information item being one of an email message, a phone message, an instant message, an alert, and a task from another application.

5. The system of claim 1, wherein the analyzer at least one of couples alerting and cost models for when to make a particular interruption of the user with considerations of the user'"'"'s current workload and applies decision-theoretic components to weigh cost of interrupting the user versus benefits of notifying the user at a given time.

6. The system of claim 1, wherein the analyzer applies bounded deferral policies to decide whether to notify users of interruptions that are potential candidates for gaining the user'"'"'s attention in the present or near future.

7. The system of claim 1, wherein the analyzer employs models that learn patterns including one or more of the user'"'"'s interests, the user'"'"'s attention modes at given times, and the user'"'"'s preference for a particular type of message.

8. The system of claim 7, wherein the models include at least one of Hidden Markov Models, Bayesian networks and other probabilistic graphical models, naï

ve Bayesian classifiers, Support Vector Machines (SVMs), neural networks, and logical rules models.

9. The system of claim 8, wherein the models receive data from one or more of a plurality of local and remote data sources including at least one of a cell phone, a microphone, a Global Positioning System (GPS), an electronic calendar, a vision monitoring system, a desktop computer, and a web site.

10. The system of claim 1, wherein the analyzer calculates a first measure of cost associated with delaying deliverance of the information item until the user'"'"'s availability is changed.

11. The system of claim 10, wherein the deferral component compares the first measure of cost with a second measure of cost associated with cost of delaying deliverance of the information item at a specified time and delivers the information item to the user based at least in part upon the comparison.

12. A method that facilitates sending notifications to a user, comprising:

utilizing a first computer system to monitor the user'"'"'s activities over time;

utilizing a second computer system to construct at least one model from the activities to determine a predicted busyness state for the user; and

utilizing a third computer system to automatically defer a received information item based at least in part on the predicted busyness state of the user and a cost model, wherein a maximal deferral time is determined based at least in part on one or more of an exact or approximate minimization of a net expected cost to the user both of delay and of interruption, the net expected cost of delay and the net expected cost of interruption scaled to a substantially similar scale, the maximal deferral time, t, being determined via one of an ideal or approximate minimization of the following formula;

W(t)=(c+ƒ

(t))g(t)+∫

₀^t−

g′

(s)ƒ

(s)ds,

where W(t) is a total expected cost;

g(t) describes a probability that the user will be busy after waiting up until the maximal deferral time t;

ƒ

(t) is a loss in the value of seeing the information item, relative to an initial value of reviewing the information item when it arrives, when reviewing the information item at the maximal deferral time t;

ƒ

(s) is a loss of value of the information item at time s that the user might become available before the maximal deferral time t under uncertainty; and

c is a cost of interruption if the user is busy;

wherein the first, second, and third computer systems may be the same or separate computer systems.

13. The method of claim 12, further comprising constructing at least one of a priorities model, a policy model, a bounded-deferral model, and a derivative model to determine the predicted busyness state for the user.

14. The method of claim 12, further comprising analyzing at least one of a session input, an attentional load factor, a user status, a threshold control, and a preference setting to determine the predicted busyness state of the user.

15. The method of claim 12, further comprising deferring interruption of the user with at least one of potentially time urgent messages, inquiries, and communications, including at least one of the relay of email messaging, instant messaging, alerts, incoming telephone calls, utterances sent via a push-to-talk communication system, error messages, inquiries to the user from autonomous systems, and advice to the user from autonomous systems.