Time-centric training, inference and user interface for personalized media program guides

• US 7,757,250 B1
• Filed: 04/04/2001
• Issued: 07/13/2010
• Est. Priority Date: 04/04/2001
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: Horvitz, Eric J., Ozer, Stuart, Kadie, Carl M.
• Primary Examiner(s): Saltarelli; Dominic D
• Application Number:
  US09/825,820
• Time in Patent Office: 3,387 Days
• Field of Search: 725/46, 725/47, 725 9- 15
• US Class Current: 725/35
• CPC Class Codes: H04H 60/37 for identifying segments of broadcast information, e.g. scenes or extracting programme ID
• CPC Class Codes: H04H 60/40 for identifying broadcast time
• CPC Class Codes: H04H 60/64 for providing detail information
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/44222 Analytics of user selections, e.g. selection of programs or purchase activity monitoring of user selections in data processing systems G06F11/34; arrangements for monitoring the user's behaviour or opinions in broadcast systems H04H60/33
• CPC Class Codes: H04N 21/4532 involving end-user characteristics, e.g. viewer profile, preferences monitoring of user activities for profile generation for accessing a video database G06F16/739; user profiles in network data switching protocols H04L67/306; processing of user preferences or user profiles in wireless networks H04W8/18
• CPC Class Codes: H04N 21/4662 characterized by learning algorithms
• CPC Class Codes: H04N 21/4668 for recommending content, e.g. movies
• CPC Class Codes: H04N 21/4756 for rating content, e.g. scoring a recommended movie
• CPC Class Codes: H04N 7/17318 Direct or substantially direct transmission and handling of requests

Abstract

The present invention is related to a system and method of considering time segments or intervals in a collaborative filtering model. The present invention extends collaborative filtering approaches by integrating considerations of temporality into the training and/or vote input associated with the usage of collaborative filtering models. The present invention also applies filtering to the output with temporal models, so as to view a most appropriate subset of recommended content, centering on content that may be available at a target time. The present invention applies time to a collaborative filtering model by allowing weight to be associated with selections within a current time segment, selections historically watched within the current time segment by the user and selections historically watched within the current time segment by a large group of users.

Claims

1. A system for ranking items in a selectable information list received from an information delivery system, comprising:

a database system that logs selections of information viewed by local users of the information delivery system and tags each of the logged selections of information with a corresponding time subinterval from a plurality of time subintervals that relates to a respective viewing of the selected information;

a filtering component that forms a temporally filtered reviewed items list that includes a subset of the logged selections of information viewed by the local users, the subset chosen to incorporate the logged selections tagged with a particular one of the plurality of time subintervals that includes a target time period for providing a recommendation, the temporally filtered reviewed items list provides implicit evidence of content preferences associated with a likely subset of the local users that employs the information delivery system during the particular one of the plurality of time subintervals;

a collaborative filtering system that infers the content preferences associated with the likely subset of the local users by utilizing the subset of the logged selections included in the temporally filtered reviewed items list as an input, and generates the recommendation specific to the inferred, likely subset of the local users based at least in part on the inferred content preferences and information obtained from a plurality of global users related to the particular one of the plurality of time subintervals, wherein the filtering component comprises a popularity filter that selects a recommendation based, at least in part, by multiplying a collaborative filter score of a recommendation by the probability that the user does not know of the recommendation; and

a user interface that displays the recommendation.

2. The system of claim 1, a selection is logged if the selection is viewed for a dwell time that exceeds a predetermined threshold.

3. The system of claim 2, the collaborative filtering system assigns a positive vote to logged selections that are viewed for a dwell time that exceeds a predetermined threshold.

4. The system of claim 1, a selection is logged if the selection is briefly viewed and jumped away to another selection.

5. The system of claim 4, the collaborative filtering system assigns a negative vote to logged selections that are viewed briefly and jumped away to another selection.

6. The system of claim 1, the viewed information is time stamped by event type and the collaborative filtering system is based on a single collaborative filtering model that is trained according to time subintervals that the information has been viewed.

7. The system of claim 1, the collaborative filtering system is based on a plurality of separate collaborative filtering models, each collaborative filtering model is trained with the information from a particular time subinterval of temporal history that has been viewed.

8. The system of claim 1, the collaborative filtering system provides in real-time a selectable recommendation list ordered by estimated degree of preference that a user has for each item.

9. The system of claim 8, the collaborative filtering system receives attributes of at least one user of the system and utilizes these attributes in providing the selectable recommendation list.

10. The system of claim 9, the collaborative filtering system receives attributes of other systems and utilizes these attributes in providing a globally ranked recommendation list to a cluster of systems based on the temporal viewing history of the systems of the cluster.

11. The system of claim 8, the collaborative filtering system receives the temporally filtered reviewed items list that has been further filtered by the filtering component and generates a new recommendation according to the preferences provided by the further filtered reviewed items list.

12. The system of claim 11, the filtering component comprising at least one of a live show selection filter, a pattern search engine or an adding items of interest and update component.

13. The system of claim 1, further comprising a user interface that allows a user to selectively filter the temporally filtered reviewed items list used as the input for the collaborative filtering system.

14. The system of claim 13, the user interface allows a user to request a time period for reviewing information from a selectable recommendation list wherein the collaborative filtering system supplies the selections for the time period requested based on the tags associated with the selections within a similar time interval covering the time period.

15. The system of claim 13, the user interface receives the temporally filtered reviewed items list, allows a user to modify the temporally filtered reviewed items list, and inputs the modified reviewed items list as updated preferences into the collaborative filtering system, such that a new recommendation list can be generated based on the updated preferences.

16. The system of claim 1, the information being multimedia.

17. A system for ranking items in a selectable information list received from an information delivery system, comprising:

means for logging selections of information viewed by local users of the information delivery system and temporal history related to time segments within a day that correspond to the viewing of the selected information, the selections of information logged for a plurality of days;

means for training a plurality of separate collaborative filtering models, each with information from a corresponding, respective time segment within a day that has been viewed by the local users and disparate logged temporal history that has been viewed by a plurality of global users;

means for inferring content preferences associated with a likely subset of the local users that employs the information delivery system during a particular time segment within a day utilizing a respective one of the collaborative filtering models corresponding to a target time period to provide a recommendation;

means for generating the recommendation specific to the inferred, likely subset of the local users based at least in part on the inferred content preferences and information obtained from a plurality of global users related to the particular time segment within a day, wherein generating the recommendation further comprises a popularity filter that selects a recommendation based, at least in part, by multiplying a collaborative filter score of a recommendation by the probability that the user does not know of the recommendation;

means for automatically broadening to include at least one additional time segment within a day when the recommendation yielded from the particular time segment within a day covering the target time period is inadequate; and

means for displaying the recommendation.