COMBINING PREDICTIVE MODELS OF FORGETTING, RELEVANCE, AND COST OF INTERRUPTION TO GUIDE AUTOMATED REMINDING

• US 20090327169A1
• Filed: 06/27/2008
• Published: 12/31/2009
• Est. Priority Date: 06/27/2008
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric, Kamar, Semiha Ece
• Granted Patent: US 7,996,338 B2
• Time in Patent Office: Days
• US Class Current: 706/11
• CPC Class Codes: G06Q 10/109 Time management, e.g. calendars, reminders, meetings or time accounting

Abstract

The claimed matter provides systems and/or techniques that develop or use predictive models of human forgetting to effectuate automated reminding. The system includes the use of predictive models that infer the probability that aspects of items will be forgotten, models that evaluate the relevance of recalling aspects of items in different settings, based on contextual information related to user attributes associated with the items, and models of the context-sensitive cost of interrupting users with reminders. The system can combine the probability of users forgetting aspects of an item with an assessed cost of forgetting those aspects to ascertain expected costs for not being reminded about events, compare expected costs for not being reminded with expected costs for interrupting users, and based on comparisons between expected costs for being reminded and expected costs for interrupting users regarding events, generate and deliver reminder notifications to users about items.

Claims

1. A system implemented on a machine that develops or utilizes predictive models of human memory to effectuate or facilitate automated reminding, comprising:

a component that acquires an event or other item via an interface, the component evaluates the event or other item for relevance based at least in part on a predictive model for the relevance of the event or other item based on contextual information or attributes associated with the event or other item, the component also infers a probability of the user forgetting about the event or other item, combines the probability of the user forgetting about the event or other item with a user specific cost of forgetting to ascertain an expected cost for not being reminded, the component compares the expected cost for not being reminded with an expected cost for interrupting the user, based at least in part on the comparison between the expected cost for being reminded and the expected cost for interrupting the user the component generates and delivers a reminder notification to the user.

2. The system of claim 1, the user assessment of the event or other item based at least in part on a user evaluated cost that includes a monetary valuation that the user assigns to avoiding a disruptive effect associated with receiving a reminder notification.

3. The system of claim 1, the component at least one of continuously or intermittently receives contextual evidence regarding a current state affiliated with the user.

4. The system of claim 3, where evidence about a user'"'"'s state includes user application context switches, user dwell periods within an application, periods of user activity, utilization of a mouse, a keyboard, or a history of the user'"'"'s computational activity or inactivity.

5. The system of claim 3, where evidence includes a user'"'"'s location, velocity, or proximity to people, resources, or structures.

6. The system of claim 3, the evidence about state including data associated with one or more events or items, the one or more events or items including appointments, conversations, discussions that the user is one or more of currently involved in, will be involved in, or has been involved in, or devices utilized by the user in one or more of taking part in the one or more events, has been employed to take part in the one or more events, or will be employed to take part in the one or more events.

7. The system of claim 1, the component determines an expected cost of interruption (ECI) based at least in part on a combination of a probability distribution of user interruptibility with costs assigned by a user for each state of interruptibility.

8. The system of claim 1, the component categorizes events into one or more priority levels based at least in part on a probabilistic model.

9. The system of claim 1, the component based at least in part on a history of high-level observations, low-level observations, or contextual information establishes a library of cases or applies Bayesian or other statistical machine learning techniques to construct a personalized predictive model that infers the cost of interrupting the user in a context.

10. The system of claim 1, the component predicts a probability of the relevance of an item or event, or surrogate for relevance that includes the probability that a user will attend an event based at least in part on properties of the item that includes one or more of the event time or date, duration, location, subject, number of invitees, or a repetition schedule associated with the event.

11. The system of claim 10, the component further predicts the probability of the user attending the event based on one or more hierarchical relationships associated with an invitee to the event or the organizer of the event, the one or more hierarchical relationships dynamically established by the component.

12. The system of claim 1, component based at least in part on the comparison between the cost for being reminded and the cost for interrupting the user generates a prognosis associated with live traffic conditions, travel routes, or travel duration associated with attending the event.

13. A machine implemented method that develops or utilizes predictive models of human forgetting and/or remembering to effectuate or facilitate automated reminding, comprising:

evaluating an item for relevance based on a user assessment of the item, contextual information associated with the user, or attributes associated with the item;

predicting a probability of the user forgetting about the item;

aggregating the probability of the user forgetting about the item with a user specific cost to ascertain a cost for not being reminded about the item;

contrasting the cost for not being reminded about the item with a cost for interrupting the user about the item; and

based at least in part on the contrasting, disseminating a reminder notification about the item to the user.

14. The method of claim 13, further comprising persisting feedback from the user and recorded in at least one of real-time or offline about whether the reminder notification was at least one of missing when needed, valuable, or inappropriate, where the reminder notification is indicated as at least one of valuable or inappropriate further soliciting from the user a reason that can include one or more of the item was not forgotten, too little detail was disseminated with the reminder notification, too much detail was associated with the reminder notification, the item was irrelevant to the user, or timing of receipt of the reminder notification was inappropriate.

15. The method of claim 13, further comprising determining an appropriate time to interrupt the user and based on the appropriate time for disseminating the reminder notification about the item to the user.

16. The method of claim 13, further comprising utilizing one or more of live traffic conditions, determined travel routes, or expected travel durations to determine an appropriate time to interrupt the user and based at least in part on the appropriate time, disseminating the reminder notification about the item to the user, the reminder notification including a proposed departure time, a latest departure time, the determined travel routes, or the expected travel duration.

17. The method of claim 13, further comprising determining a probability of the user attending an event, the probability of the user attending the event based on an event time or date, an event duration, a repetition cycle associated with the event, a unique location associated with the event, or an involvement the user has with the event.

18. A system that utilizes predictive models of human memory to effectuate automated reminding, comprising:

means for assessing an event for relevance based on a user assessment of the item, contextual information associated with the user, or attributes associated with the item;

means for inferring a probability of the user forgetting about one or more aspects of the item;

means for combining the probability of the user forgetting about the item with a user specific cost to ascertain a cost for not being reminded about one or more aspects of the item;

means for contrasting the cost for not being reminded about one or more aspects of the item with a cost for interrupting the user about the event; and

means for disseminating a reminder notification about the event to the user.

19. The system of claim 18, further comprising means for obtaining a user evaluated cost associated with the item, the user evaluated cost includes an avoidance cost that the user assigns to the item.

20. The system of claim 18, further comprising means for timing the reminder notification based at least in part on a means for determining live traffic conditions, the means for determining live traffic conditions provides dynamic travel route updates, or an expected travel duration, the expected travel duration utilized by the means for disseminating to disseminate the reminder notification.