Methods and architecture for learning and reasoning in support of context-sensitive reminding, informing, and service facilitation

• US 7,797,267 B2
• Filed: 06/30/2006
• Issued: 09/14/2010
• Est. Priority Date: 06/30/2006
• Status: Active Grant

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J.
• Primary Examiner(s): Sparks; Donald
• Assistant Examiner(s): Wong; Lut
• Application Number:
  US11/428,225
• Publication Number:
  US 20080005055A1
• Time in Patent Office: 1,537 Days
• Field of Search: 706/46, 706/52, 706/21
• US Class Current: 706/46
• CPC Class Codes: G06Q 10/04 Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem" market predictions or forecasting for commercial activities G06Q30/0202
• CPC Class Codes: G06Q 30/06 Buying, selling or leasing transactions

Abstract

Methods and architectures for context-sensitive reminding and service facilitating are disclosed. The architectures monitor user context and activity, senses or infers relevant reminders, goals, such as those that come from a growing need of the user that should be fulfilled, and computes best reminders, and recommend plans on fulfilling need(s) in an optimum way. Statistical models of a user'"'"'s knowledge and recall in different settings may be employed. Facilities, services, and merchants can be identified along a route that the user can take, and cost-benefit analysis is performed for determining which merchant(s) to select to fulfill the need(s). Routes may be created as opportunistic modifications of trips underway. Merchants can respond back with offers of sale to the user for all available needed items, and the user can respond with acceptance or denial of the offers. Merchants can also respond in a bidding fashion in order to gain user'"'"'s patronage.

Claims

1. A computer-implemented system for providing an opportunistic reminder, comprising:

a context component that determines a context of an entity; and

a processor executing an indication component that employs a probabilistic model to provide a reminder of a fact to the entity based at least in part upon the context and an expected value of providing the reminder, wherein;

the indication component selectively outputs the reminder based upon a cost-benefit evaluation of whether to provide the reminder, and

the expected value of providing the reminder is based at least in part on a value of the entity being informed of the fact within the context, a likelihood that the entity will recall the fact within a time frame and a cost of interruption to inform the entity of the fact.

2. The system of claim 1, the indication component further comprising a relevance component that reasons about a likelihood of a fact k being relevant to the entity by evaluating a probability that the fact k is relevant given the context, p(relevant(k)|context).

3. The system of claim 1, the indication component further comprising a knowledge component that evaluates whether the entity at least one of knows a fact k and will recall the fact k within a time frame, the knowledge component computes a probability that the entity knows the fact k given the context, p(knows(k)|context).

4. The system of claim 1, the indication component further comprising a valuation component that analyzes the value of the entity being informed of the fact within the context, the value comprising a value of the entity knowing of the fact.

5. The system of claim 4, the valuation component evaluates a value of being informed about the fact, k, by analyzing a difference between a value of knowing versus not knowing k within the context, the valuation component computes Utility(know(k)|context)−

Utility(not know(k)|context).

6. The system of claim 1, the indication component further comprising an integrative component that analyzes an expected value of reminding the entity of the fact, k.

7. The system of claim 6, the integrative component evaluates a probability that the entity does not know the fact k, given a probabilistic combination of a probability that the fact k is relevant in the context and a probability that the entity would know the fact k, given that the fact k is relevant, p(not know(k)|context)=1=(know(k)|relevant(k), context)*p(relevant(k)|context).

8. The system of claim 6, the integrative component computes the expected value of informing the entity about the fact k in the context under uncertainty, whether the entity knows the fact k or not ahead of time, by evaluating eu(inform(k)|context)p(not know(k)|context)*[utility(know(k))−

utility(not know(k)|context)]−

cost(interruption(inform(k))|context), where eu is the expected value.

9. The system of claim 1, the indication component employs user specified high-level preferences and goals that define classes of knowledge as being at least one of relevant and having a likelihood above a threshold of being relevant.

10. The system of claim 1, the indication component generates opportunistic reminders related to user knowledge of at least one of the fact comprising a task desired, a need, a user'"'"'s lack of an amount of an item, and a potential opportunity.

11. The system of claim 1, wherein:

the entity is a person;

the indication component evaluates a suspended task indicated and performed by the person; and

the reminder reminds the person that the task is suspended to enable recovery from interruption associated with the suspended task.

12. The system of claim 1, the context being at least one of a location, information associated with a suspended task, a need, a status of expendable resources, and proximal businesses, services, terrain, and recreational facilities.

13. A computer-implemented method of selectively providing opportunistic reminders, comprising:

evaluating a context of a user;

analyzing whether to output a reminder of information to the user based upon a cost-benefit evaluation of providing the reminder and upon considerations of an expected value of providing the reminder and cost of informing the user in view of the context; and

selectively outputting the reminder based upon the analysis,

wherein the expected value of providing the reminder is based at least in part on a value of the user being informed of the information within the context, a likelihood that the user will recall the information within a time frame and a cost of interruption to provide the user the reminder.

14. The method of claim 13, evaluating the context further comprises analyzing at least one of a location of the user, a preference of the user, a calendar of the user, a need of the user, a time, a day, a month, a forthcoming meeting, a status of expendable resources, and proximal businesses, services, terrain, and recreational facilities.

15. The method of claim 13, further comprising inferring a likelihood of a fact being relevant to the user.

16. The method of claim 13, wherein the information comprises a fact, the method further comprising inferring a likelihood that the user knows about the fact and will recall the fact within the time frame.

17. The method of claim 13, wherein the information comprises a fact and the value of the user being informed of the information comprises a value of the user knowing the fact, the method further comprising evaluating a value of the user knowing by way of being reminded about the fact.

18. The method of claim 13, farther comprising selectively outputting the reminder based upon explicit and learned goals and preferences of the user.

19. The method of claim 13, further comprising selectively outputting the reminder by maximizing an expected utility associated with the user.

20. A computer-executable system for generating reminders, the system comprising:

means for determining a context of a user;

means for inferring, by the at least one processor, a first likelihood of a fact being relevant to the user in view of the context;

means for determining, by the at least one processor, a second likelihood that the user at least one of knows the fact and will recall the fact within a time frame given the context;

means for analyzing, by the at least one processor, a utility of the user knowing the fact given the context;

means for evaluating, by the at least one processor, cost of informing the user in view of the context; and

means for selectively outputting the reminder based upon a cost-benefit evaluation of whether to provide the reminder based on the utility of providing the reminder, the second likelihood and the cost of providing the reminder.

21. A computer-implemented system for providing an opportunistic reminder, comprising:

a context component that determines a context of an entity; and

a processor executing an indication component that employs a probabilistic model to selectively provide a reminder to the entity based at least in part upon the context, the indication component further comprising a valuation component that analyzes a value of the entity at least one of knowing and being reminded of a fact, the valuation component evaluates a value of being informed about the fact k by analyzing a difference between a value of knowing versus not knowing k within the context, the valuation component computes Utility(know(k)|context)−

Utility(not know(k)|context).

22. A computer-implemented system for providing an opportunistic reminder, comprising:

a context component that determines a context of an entity; and

a processor executing an indication component that selectively provides a reminder to the entity based at least in part upon the context, the indication component further comprising;

a valuation component computes a first value based on the entity knowing the fact within the context;

the valuation component computes a second value based on the entity not knowing the fact within the context; and

the valuation component computes a third value based on cost of interruption to the entity;

the valuation component determining a resulting value associated with a utility of reminding the entity of the fact based at least in part on the first value, the second value and the third value.

23. The computer-implemented system of claim 22, wherein the valuation component determines the resulting value associated with a utility of reminding the entity of the fact by analyzing a difference between the first value and the second value.