Building and using predictive models of current and future surprises

• US 8,706,651 B2
• Filed: 04/03/2009
• Issued: 04/22/2014
• Est. Priority Date: 11/16/2004
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J.
• Primary Examiner(s): Chaki, Kakali
• Assistant Examiner(s): Seck, Ababacar
• Application Number:
  US12/418,226
• Publication Number:
  US 20110071964A1
• Time in Patent Office: 1,845 Days
• Field of Search: None
• US Class Current: 706/12
• CPC Class Codes: G06F 18/214 Generating training patterns; Bootstrap methods, e.g. bagging or boosting
• CPC Class Codes: G06F 18/2411 based on the proximity to a decision surface, e.g. support vector machines
• CPC Class Codes: G06N 20/00 Machine learning
• CPC Class Codes: G06N 5/025 Extracting rules from data
• CPC Class Codes: G06N 7/01 Probabilistic graphical models, e.g. probabilistic networks
• CPC Class Codes: G08G 1/0104 Measuring and analyzing of parameters relative to traffic conditions

Abstract

Methods are described for identifying events that would be considered surprising by people and identifying how and when to transmit information to a user about situations that they would likely find surprising. Additionally, the methods of identifying surprising situations can be used to build a case library of surprising events, joined with a set of observations before the surprising events occurred. Statistical machine learning methods can be applied with data from the case library to build models that can predict when a user will likely be surprised at future times. One or more models of context-sensitive expectations of people, a view of the current world, and methods for recording streams or events before surprises occur, and for building predictive models from a case library of surprises and such historical observations can be employed. The models of current and future surprises can be coupled with display and alerting machinery.

Claims

1. A system for providing notifications of events, the system comprising:

a memory and a processor that are respectively configured to store and execute computer-executable instructions, including instructions that are organized into;

an interface component that receives a dataset having at least some contextual data, and that provides the received dataset to a predictive model component, wherein the at least some contextual data is at least potentially associated with an event;

the predictive model component, wherein the predictive model component generates a first prediction for an event based at least on a portion of the dataset, and generates a second prediction for the event based at least on the portion of the dataset or on another portion of the dataset, wherein the second prediction is expected to have a greater accuracy than the first prediction; and

a difference analyzer component that selectively outputs an indication of the second prediction based at least on whether the first prediction is different than the second prediction, wherein the first prediction is based on a user expectancy model that is designed to emulate predictive capabilities of a user.

2. The system of claim 1, wherein the first prediction and the second prediction include probabilities of correctness, and wherein the difference analyzer component selectively outputs the indication based also on the probabilities of correctness.

3. The system of claim 1, wherein the indication is representative of a likelihood that the user would be surprised by the event.

4. The system of claim 1, wherein the user expectancy model is designed to emulate whether the user would predict the event.

5. The system of claim 1, wherein the user expectancy model is based on analysis of variables of the dataset and interrelationships between the variables, and wherein the user expectancy model corresponds to human modes of predicting events.

6. The system of claim 5, wherein:

the dataset also includes historical data for other events similar to the event;

the predictive model component generates the first prediction based at least on a portion of the contextual data and a portion of the historical data; and

the predictive model component generates the second prediction based at least on the portion of the contextual data or another portion of the contextual data and on the portion of the historical data or another portion of the historical data.

7. The system of claim 1, wherein the computer-executable instructions include further instructions that are organized into:

a utility analyzer that calculates a measure of utility associated with a potential outputting of the indication, wherein the difference analyzer component outputs the indication if the measure of utility is above a threshold.

8. The system of claim 1, wherein the computer-executable instructions include further instructions that are organized into:

an alerting component that utilizes alerting rules to determine a manner in which the indication is to be output by the difference analyzer component.

9. The system of claim 1, wherein:

the first prediction is a prediction of a roadway traffic congestion condition and is based at least on typical traffic congestion conditions for a current time of day;

the first prediction and the second prediction are different; and

the indication represents that the event is atypical for the current time of day.

10. The system of claim 1, wherein the indication represents a prediction that a traffic congestion condition will happen at a future time.

11. A method of providing user notifications for predicted events, the method comprising:

receiving a first predictive model, wherein the first predictive model is based on a user expectancy model designed to reflect whether a user would predict the event;

receiving a second predictive model that is different than the first predictive model;

receiving contextual data that is at least potentially associated with an event;

with at least one processor, applying the first predictive model to at least a portion of the contextual data to generate a first prediction for the event;

with the at least one processor, applying the second predictive model to at least the portion of the contextual data or to at least another portion of the contextual data to generate a second prediction for the event;

determining that the first prediction is different than the second prediction; and

in response to the determination, selectively outputting an indication that the event is predicted.

12. The method of claim 11, wherein:

the first prediction is also based at least on a current time and/or day;

the first prediction and the second prediction are different; and

the indication represents that the event is atypical for the current time and/or day.

13. The method of claim 11, wherein:

the first prediction is a prediction of a roadway traffic congestion condition and is also based at least on typical traffic congestion conditions for a current time of a day;

the first prediction and the second prediction are different; and

the indication represents that the event is atypical for the current time of the day.

14. The method of claim 13, wherein the indication also represents a prediction of atypically less than typical traffic congestion along a route.

15. The method of claim 13, wherein the second predictive model includes one or more statistical models that employ Bayesian networks, dynamic Bayesian networks, continuous time Bayesian networks, support vector machines, neural network models, Hidden Markov Models, Markov decision processes, and/or particle filtering.

16. The method of claim 11, wherein the first and second predictions are predictions of roadway traffic conditions at a future time.

17. The method of claim 11, further comprising:

calculating a measure of utility associated with outputting an indication for the predicted event; and

outputting the indication if the measure of utility associated therewith is above a threshold.

18. The method of claim 11, wherein:

the first predictive model also corresponds to human modes of predicting events;

the second predictive model is expected to be more accurate than the user expectancy model; and

the indication is representative of a likelihood that the user would be surprised by the event.

19. A computer-readable storage device comprising computer-executable instructions for, when executed by at least one processor, implementing a predictive system, the computer-executable instructions for executing a method comprising:

receiving a user expectancy model that is designed to emulate predictive capabilities of a user of the predictive system;

receiving a second model that is expected to be more accurate than the user expectancy model;

receiving contextual data that is at least potentially associated with a traffic congestion event;

generating a first prediction for the traffic congestion event with the user expectancy model and at least some contextual data;

generating a second prediction for the traffic congestion event with the second model and the at least some contextual data;

determining that the first prediction is different than the second prediction; and

in response to the determination, providing information to the user regarding the traffic congestion condition based on the second prediction.

20. The computer-readable storage device of claim 19, wherein:

the first prediction is a prediction of traffic congestion along a route and is based at least on typical traffic congestion along the route at a time of day; and

the indication also represents a prediction of atypically less than typical traffic congestion along the route for the time of day.