METHOD AND APPARATUS FOR AUTOMATICALLY INCORPORATING HYPOTHETICAL CONTEXT INFORMATION INTO RECOMMENDATION QUERIES

• US 20090193099A1
• Filed: 01/29/2008
• Published: 07/30/2009
• Est. Priority Date: 01/29/2008
• Status: Active Grant

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Palo Alto Research Center, Inc. (Xerox Holdings Corp.)
• Inventors: Partridge, Kurt E., Ducheneaut, Nicolas B., Price, Robert R.
• Granted Patent: US 7,904,530 B2
• Time in Patent Office: Days
• US Class Current: 709/219
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: H04W 4/029 Location-based management or tracking services

Abstract

A system facilitates automatically determining the hypothetical context information or the distribution of hypothetical contexts. During operation, the system receives a request from a user for one or more recommendations. The system also receives a current context substantially associated with the request. The system then determines a hypothetical context for the request, wherein the hypothetical context may be determined by considering several sources of information, including but not limited to the current context, past contexts, and relationships between the current context and past contexts. Next, the system determines one or more recommendations for the user based on the hypothetical context. Finally, the system returns the one or more recommendations to the user.

Claims

1. A method for automatically incorporating hypothetical context information into recommendations, the method comprising:

receiving a request from a user for one or more recommendations;

receiving a current context substantially associated with the request;

determining a hypothetical context for the request;

wherein the hypothetical context may be determined from the current context;

wherein the hypothetical context may comprise one or more future times and/or one or more future locations associated with the user; and

wherein a respective hypothetical context may contain a joint distribution over one or more contextual variables which can have discrete values, continuous values, and/or a combination of discrete and continuous values;

determining one or more recommendations for the user based on the hypothetical context; and

returning the one or more recommendations to the user.

2. The method of claim 1, wherein determining the hypothetical context comprises:

constructing a mapping function;

applying the mapping function to the current context; and

producing a hypothetical context as an output of the mapping function.

3. The method of claim 2, wherein constructing the mapping function comprises determining the mapping function by machine learning methods based on one or more of the following data sources:

prior contexts associated with prior user requests;

prior hypothetical contexts which are derived in association with prior user requests;

the mapping relationship between the prior contexts and corresponding prior hypothetical contexts associated with the same prior user requests; and

user interaction associated with prior user requests for recommendation.

4. The method of claim 1, wherein determining a respective hypothetical context can involve:

estimating a distribution over time;

estimating a distribution over locations;

determining a distribution over weather conditions for the distribution over locations and the distribution over time;

determining a distribution over context-related preferences; and

/or

determining a distribution over conditions related to the hypothetical context that may be inferred from publicly-available data about the distribution over other contextual variables.

5. The method of claim 4, wherein estimating the distribution over time involves:

examining one or more prior user requests associated with the location of the current context;

determining the context associated with each prior user request;

determining the hypothetical context which has been derived in association with each prior user request; and

determining a mapping from the current context to the time distribution of the corresponding hypothetical context, wherein determining the mapping can involve producing a time distribution based on the time associated with the current context and/or producing a time distribution based on the time associated with a calendar entry;

constructing a mapping function based on the determined mapping; and

applying the mapping function to the current context to determine the distribution over time.

6. The method of claim 4, wherein estimating the distribution over locations can involve:

collecting a set of candidate locations from location distributions that the user has previously specified and/or locations that the user has previously visited, optionally excluding;

candidate locations that have not been visited more than a predetermined number of times;

location distributions that have not been specified more than a predetermined number of times;

location distributions that were not specified during a time interval derived from the time associated with the current context; and

/or

candidate locations that were not visited during one or more time periods corresponding to the time distribution associated with the hypothetical context; and

determining a distribution over locations from the set of candidate locations or one or more statistics of the set of candidate locations.

7. The method of claim 6, wherein the statistics of the set of candidate locations are determined in part from the type of venue associated with each candidate location;

and

wherein the distribution over locations can include locations not in the set of candidate locations but related to the type of venue of at least one candidate location.

8. The method of claim 4, wherein determining the hypothetical context involves:

retrieving from a calendar a context whose associated time period substantially matches the time distribution of the hypothetical context; and

setting other variables of the hypothetical context to the variables of the context retrieved from the calendar.

9. The method of claim 4, wherein determining the weather condition distribution can include performing a lookup to determine a distribution over forecasted weather conditions corresponding to the distribution of location and time.

10. The method of claim 1, further comprising saving the hypothetical context to a calendar to facilitate subsequent hypothetical context determinations.

11. The method of claim 1, further comprising:

presenting the hypothetical context to the user;

receiving a change to the hypothetical context from the user; and

applying the change to the hypothetical context.

12. A computer-readable storage medium storing instructions that when executed by a computer cause the computer to perform a method for automatically incorporating hypothetical context information into recommendations, the method comprising:

receiving a request from a user for one or more recommendations;

receiving a current context substantially associated with the request;

determining a hypothetical context for the request;

wherein the hypothetical context may be determined from the current context;

wherein the hypothetical context may comprise one or more future times and/or one or more future locations associated with the user; and

wherein a respective hypothetical context may contain a joint distribution over one or more contextual variables which can have discrete values, continuous values, and/or a combination of discrete and continuous values;

determining one or more recommendations for the user based on the hypothetical context; and

returning the one or more recommendations to the user.

13. The computer-readable storage medium of claim 12, wherein determining the hypothetical context comprises:

constructing a mapping function;

applying the mapping function to the current context; and

producing a hypothetical context as an output of the mapping function.

14. The computer-readable storage medium of claim 13, wherein constructing the mapping function comprises determining the mapping function by machine learning methods based on one or more of the following data sources:

prior contexts associated with prior user requests;

prior hypothetical contexts which are derived in association with prior user requests;

the mapping relationship between the prior contexts and corresponding prior hypothetical contexts associated with the same prior user requests; and

user interaction associated with prior user requests for recommendation.

15. The computer-readable storage medium of claim 12, wherein determining a respective hypothetical context can involve:

estimating a distribution over time;

estimating a distribution over locations;

determining a distribution over weather conditions for the distribution over location and the distribution over time;

determining a distribution over context-related preferences; and

/or

determining a distribution over conditions related to the hypothetical context that may be inferred from publicly-available data about the distribution over other contextual variables.

16. The computer-readable storage medium of claim 15, wherein estimating the distribution over time involves:

examining one or more prior user requests associated with the location of the current context;

determining the context associated with each prior user request;

determining the hypothetical context which has derived in association with each prior user request; and

determining a mapping from the current context to the time distribution of the corresponding hypothetical context, wherein determining the mapping can involve producing a time distribution based on the time associated with the current context and/or producing a time distribution based on the time associated with a calendar entry;

constructing a mapping function based on the determined mapping; and

applying the mapping function to the current context to determine the distribution over time.

17. The computer-readable storage medium of claim 15, wherein estimating the distribution over locations can involve:

collecting a set of candidate locations from location distributions that the user has previously specified and/or locations that the user has previously visited, optionally excluding;

candidate locations that have not been visited more than a predetermined number of times;

locations distributions that have not been specified more than a predetermined number of times;

locations distributions that were not specified during a time interval

derived from the time associated with the current context; and

/or

candidate locations that were not visited during one or more time periods corresponding to the time distribution associated with the hypothetical context; and

determining a distribution over locations from the set of candidate locations or one or more statistics of the set of candidate locations.

18. The computer-readable storage medium of claim 17, wherein the statistics of the set of candidate locations are determined in part from the type of venue associated with each candidate location;

and

wherein the distribution over locations can include locations not in the set of candidate locations but related to the type of venue of at least one candidate location.

19. The computer-readable storage medium of claim 15, wherein determining the hypothetical context involves:

retrieving from a calendar a context whose associated time period substantially matches the time distribution of the hypothetical context; and

setting other variables of the hypothetical context to the variables of the context retrieved from the calendar.

20. The computer-readable storage medium of claim 15, wherein determining the weather condition distribution can include:

performing a lookup to determine a distribution over forecasted weather condition corresponding to the distribution of location and time

21. The computer-readable storage medium of claim 12, wherein the method further comprises saving the hypothetical context to a calendar to facilitate subsequent hypothetical context determinations.

22. The computer-readable storage medium of claim 12, wherein the method further comprises:

presenting the hypothetical context to the user;

receiving a change to the hypothetical context from the user; and

applying the change to the hypothetical context.

23. An apparatus for automatically incorporating hypothetical context information into recommendations, the apparatus comprising:

a request receiving mechanism configured to receive a request from a user for one or more recommendations;

a context receiving mechanism configured to receive a current context substantially associated with the request;

a first determination mechanism configured to determine a hypothetical context for the request;

wherein the hypothetical context may be determined from the current context;

wherein the hypothetical context may comprise one or more future times and/or one or more future locations associated with the user; and

wherein a respective hypothetical context may contain a joint distribution over one or more contextual variables which can have discrete values, continuous values, and/or a combination of discrete and continuous values;

a second determination mechanism configured to determine one or more recommendations for the user based on the hypothetical context; and

a display mechanism configured to return the one or more recommendations to the user.