Intelligent system and methods of recommending media content items based on user preferences

• US 7,840,986 B2
• Filed: 12/14/2000
• Issued: 11/23/2010
• Est. Priority Date: 12/21/1999
• Status: Active Grant

• Current Assignee: TiVo Solutions Inc. (Adeia Inc.)
• Original Assignee: TiVo Inc. (Adeia Inc.)
• Inventors: Stam, Wijnand Van, Ali, Kamal
• Primary Examiner(s): Kelley; Christopher
• Assistant Examiner(s): Andramuno; Franklin S
• Application Number:
  US10/168,808
• Publication Number:
  US 20020199186A1
• Time in Patent Office: 3,631 Days
• Field of Search: 725/105, 725/46, 725/32, 725/34, 725/35
• US Class Current: 725/105
• CPC Class Codes: G06F 16/735 Filtering based on additional data, e.g. user or group profiles
• CPC Class Codes: G06F 16/7867 using information manually generated, e.g. tags, keywords, comments, title and artist information, manually generated time, location and usage information, user ratings
• CPC Class Codes: G11B 2220/20 Disc-shaped record carriers
• CPC Class Codes: G11B 27/105 of operating discs
• CPC Class Codes: G11B 27/34 Indicating arrangements indicating means incorporated in magazine or cassette G11B23/046 and G11B23/0875; indicating measured values in general G01D
• CPC Class Codes: H04N 21/252 Processing of multiple end-users' preferences to derive collaborative data
• CPC Class Codes: H04N 21/4147 PVR [Personal Video Recorder] H04N5/76 takes precedence ; arrangements for broadcast with accumulation-type receivers H04H20/40
• CPC Class Codes: H04N 21/4316 for displaying supplemental content in a region of the screen, e.g. an advertisement in a separate window
• CPC Class Codes: H04N 21/4334 Recording operations recording of a television signal H04N5/76; arrangements for recording or accumulating broadcast information or broadcast-related information H04H60/27
• 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 21/443 OS processes, e.g. booting an STB, implementing a Java virtual machine in an STB or power management in an STB
• 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 21/454 Content or additional data filtering, e.g. blocking advertisements
• CPC Class Codes: H04N 21/466 Learning process for intelligent management, e.g. learning user preferences for recommending movies services using the results of monitoring in broadcast systems H04H60/61
• CPC Class Codes: H04N 21/4663 involving probabilistic networks, e.g. Bayesian networks
• CPC Class Codes: H04N 21/47 End-user applications
• CPC Class Codes: H04N 21/47214 for content reservation or setting reminders; for requesting event notification, e.g. of sport results or stock market stock exchange data over packet-switching network H04L12/1804; push services over packet-switching network H04L12/1859; notification of incoming messages in packet switching networks H04L51/224
• CPC Class Codes: H04N 21/4756 for rating content, e.g. scoring a recommended movie
• CPC Class Codes: H04N 21/478 Supplemental services, e.g. displaying phone caller identification, shopping application
• CPC Class Codes: H04N 21/4826 using recommendation lists, e.g. of programs or channels sorted out according to their score
• CPC Class Codes: H04N 21/4828 : for searching program descriptors retrieval of video data G06F16/739
• CPC Class Codes: H04N 21/485 : End-user interface for client configuration
• CPC Class Codes: H04N 21/6543 : for forcing some client operations, e.g. recording remote booting in general G06F9/4416
• CPC Class Codes: H04N 21/812 : involving advertisement data advertising per se G06Q30/02
• CPC Class Codes: H04N 21/8166 : involving executable data, e.g. software arrangements for executing specific programs G06F9/44; broadcasting computer programmes in broadcast systems H04H20/91; involving the movement of software or configuration parameters H04L67/34
• CPC Class Codes: H04N 5/76 : Television signal recording
• CPC Class Codes: H04N 5/782 : on tape
• CPC Class Codes: H04N 7/173 : with two-way working, e.g. subscriber sending a programme selection signal
• CPC Class Codes: H04N 7/17318 : Direct or substantially direct transmission and handling of requests
• CPC Class Codes: H04N 7/17354 : in an intermediate station common to a plurality of user terminals

Abstract

A system and method for making program recommendations to users of a network-based video recording system utilizes expressed preferences as inputs to collaborative filtering and Bayesian predictive algorithms to rate television programs using a graphical rating system. The predictive algorithms are adaptive, improving in accuracy as more programs are rated.

Claims

1. A distributed system for predicting items for a user, the system comprising:

a client;

a server in communication with a plurality of clients, including the client, over a network connection;

device logic at the server that periodically receives a list of user-rated items from each client of the plurality of clients, the lists of user-rated items aggregated into a single aggregated list of items, the items being associated with media content, the user-rated items in the list of user-rated items from the client being rated by a user, the user-rated items in the lists of user-rated items from other clients of the plurality of clients being rated by other users;

logic at the server that filters user-rated items from the single aggregated list of items based on frequency by monitoring frequency of the user-rated items and discarding items and the items'"'"' corresponding user-ratings that do not satisfy a threshold frequency;

logic at the server that creates matrices, each matrix of the matrices corresponding to each unique pair of items from the single aggregated list of items, each matrix storing user-ratings for each item of the unique pair of items, the matrices anonymous with respect to the user and the other users;

logic at the server that computes a rating correlation between items of the unique pair of items from each matrix;

logic at the server that filters out non-significant rating correlations of unique pairs of items;

logic at the server that compiles a list of correlating items comprising for which the rating correlations were not filtered out;

logic at the server that periodically sends the list of correlating items to the client; and

logic at the client that predicts a rating for an unrated item based on the correlations provided in the list of correlating items.

2. The system of claim 1, wherein the connection between any of the plurality of clients and the server is stateless.

3. The system of claim 1, wherein the list of user-rated items received by the server from the client includes each item rated by the user and the corresponding rating assigned by the user.

4. The system of claim 3, wherein the corresponding rating assigned by the user is assigned according to a scalar system of discrete ratings.

5. The system of claim 1, wherein dimensions of each matrix of the matrices are n by n, where n equals a number of discrete user-ratings.

6. The system of claim 1, wherein the user-ratings are selected from a group −

3, −

2, −

1, 0, 1, 2, and 3, and a negative value indicates an unfavorable rating, a positive value indicates a favorable rating, and a ‘

0’

indicates a neutral rating.

7. The system of claim 5, wherein n=7.

8. The system of claim 1, wherein pairs of user-ratings are saved to the matrices by:

retrieving a pair of user-ratings for a pair of items;

determining whether a matrix storing user-ratings for the pair of items exists;

creating the matrix for the pair of items if the matrix does not exist;

locating a cell that stores the pair of user-ratings in the matrix by using the pair of user-ratings as an index to the matrix; and

incrementing the cell'"'"'s value that corresponds to the pair of user-ratings by 1.

9. The system of claim 1, wherein a correlation factor is calculated for each matrix.

10. The system of claim 9, where the correlation factor is of the range of −

1 to 1.

11. The system of claim 9, wherein a positive correlation factor indicates a positive correlation and a negative correlation factor indicates an anti-correlation.

12. The system of claim 1, wherein the logic at the server that filters non-significant correlations comprises a correlation filter that filters item pairs that correlate weakly.

13. The system of claim 12, wherein a correlation factor having a non-zero value that approaches zero indicates a weak correlation between items of a pair.

14. The system of claim 1, wherein the logic at the client that predicts a rating for an unrated item comprises a prediction engine, the prediction engine predicting a rating for an unrated item by:

receiving an unrated item;

searching for the unrated item in the list of correlating items;

creating a work list of correlation factors for all items in the list of correlating items correlating to the unrated item, that have been rated by the user;

insuring that the work list is of a predetermined, fixed length by removing most weakly correlating items when the work list exceeds the predetermined fixed length and padding the work list with correlation factors of 1 and neutral ratings when the work list is shorter than the predetermined fixed length;

multiplying a value for the user-rating and the correlation factor for each item in the work list and summing products of the value for the user-rating and the correlation factor; and

dividing a sum of the products of the value for the rating and the correlation factor by a sum of absolute values of correlation factors for each item in the work list, a resulting value constituting a predicted rating for the unrated item.

15. The system of claim 1, wherein the server comprises a single computer having multiple functional units.

16. The system of claim 1, wherein the server comprises a single server having a plurality of functional units and separate stages of the system are implemented in separate functional units.

17. The system of claim 1, wherein the server comprises a plurality of computers and separate stages of the system are implemented on separate computers, the computers being in communication with other separate computers of the server.

18. The system of claim 1, wherein the media content comprises any of network television programming, cable television programming, films, pay-per-view television programming and video.

19. The system of claim 1, the system being a component of a network-based video recording system.

20. A method of compiling a list of correlated items, the method comprising the steps of:

periodically receiving at a server a list of items rated by the user from a client of a plurality of clients over a stateless network connection, the items being associated with media content;

aggregating the list of items at the server with lists from other clients of the plurality of clients in contact with the server into a single aggregated list of items, the items in the lists from the other clients rated by other users of a plurality of users;

filtering the single aggregated list of items by monitoring frequency of user-rated items, and discarding items and corresponding user-ratings that do not satisfy a threshold frequency;

tallying user-ratings for each item of each unique pair of items from the single aggregated list of items and storing tallies in one or more pair matrices, each pair matrix corresponding to each unique pair of items, the one or more pair matrices anonymous with respect to users of the plurality of users;

computing a rating correlation between items of the unique pair of items from each pair matrix;

filtering out non-significant rating correlations of unique pairs of items;

compiling a list of correlating items comprising items for which the rating correlations were not filtered out; and

periodically sending by the server the list of correlating items to the client of the plurality of clients.

21. The method of claim 20, wherein the list of user-rated items received at the server comprises the list of user-rated items from the client of the plurality of clients, and the list of user-rated items from the client includes each item rated by the user and the corresponding rating assigned by the user.

22. The method of claim 20, wherein the user rates the items according to a scalar system of discrete ratings.

23. The method of claim 20, wherein dimensions of the one or more pair matrices are n by n, where n equals a number of discrete user-ratings.

24. The method of claim 23, wherein n=7.

25. The method of claim 20, wherein the step of tallying user-ratings for each item and storing tallies in one or more pair matrices further comprises the steps of:

retrieving a pair of user-ratings for a pair of items;

determining whether the matrix storing ratings of the pair of items exists;

creating the matrix storing the pair of items if the matrix does not exist;

locating a cell that stores the pair of user-ratings in the matrix by using the pair of user-ratings as an index to the matrix; and

incrementing the cell'"'"'s value that corresponds to the pair of user-ratings by 1.

26. The method of claim 20, wherein a correlation factor is calculated for each matrix.

27. The method of claim 26, wherein the correlation factor is of the range of −

1 to 1.

28. The method of claim 26, wherein a positive correlation factor indicates a positive correlation and a negative correlation factor constitutes an anti-correlation.

29. The method of claim 20, wherein the step of filtering out non-significant correlations comprises the step of:

filtering out correlations with a correlation factor having a non-zero value that approaches zero.

30. The method of claim 20 further comprising a client predicting a rating for an unrated item comprising:

searching for the unrated item in the list of correlating items;

creating a work list of correlation factors from all items in the list of correlating items correlating to the unrated item, that have been rated by the user;

insuring that the work list is of a predetermined, fixed length;

multiplying a value for the user-rating and the correlation factor for each item in the work list, and summing products of the value for the user-rating and the correlation factor; and

dividing the sum of the products of the value for the user-rating and the correlation factor by a sum of absolute values of correlation factors for each item in the work list, a resulting value constituting a predicted rating for the unrated item.

31. The method of claim 30, wherein the step of insuring that the work list is of a predetermined, fixed length comprises one of the steps of:

removing most weakly correlating items when the work list exceeds the predetermined fixed length; and

padding the work list with correlation factors of one and neutral ratings when the work list is shorter than the predetermined, fixed length.

32. The method of claim 20, wherein the server comprises a single server having a plurality of functional units and separate stages of the method are executed in separate functional units.

33. The method of claim 20, wherein the server comprises a plurality of computers and separate stages of the method are executed on separate computers, the computers being in communication with other separate computers of the server.

34. The method of claim 20, wherein the media content comprises any of network television programming, cable television programming, films, pay-per-view television programming, and video.

35. The method of claim 20, wherein the method is implemented within a network-based video recording system.