Intelligent system and methods of recommending media content items based on user preferences
First Claim
1. A distributed system for predicting items likely to appeal to a user, based on a client-server collaborative filtering engine, wherein ratings are predicted for said items on the client side using correlation factors downloaded from the server, said correlation factors being computed on the server side from preference profiles anonymously posted to said server from a plurality of clients, said system comprising:
- a plurality of clients;
a server side, said clients in intermittent communication with said server side over a network connection;
a list of rated items from each client, wherein said lists are periodically transmitted to said server and aggregated into a single list;
means for filtering said rated items based on frequency;
a matrix for each unique pair of items, wherein said matrix tallies ratings for each item of said pair;
means for computing a correlation between items of said pair from said matrix;
means for filtering non-significant correlations;
a list of correlating items, said list comprising a list of all significant correlations, wherein said list is periodically transmitted to at least one client from said server side; and
client-side means for predicting a rating for an unrated item based on the correlations provided in the list of correlating items.
12 Assignments
0 Petitions
Accused Products
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.
-
Citations
40 Claims
-
1. A distributed system for predicting items likely to appeal to a user, based on a client-server collaborative filtering engine, wherein ratings are predicted for said items on the client side using correlation factors downloaded from the server, said correlation factors being computed on the server side from preference profiles anonymously posted to said server from a plurality of clients, said system comprising:
-
a plurality of clients;
a server side, said clients in intermittent communication with said server side over a network connection;
a list of rated items from each client, wherein said lists are periodically transmitted to said server and aggregated into a single list;
means for filtering said rated items based on frequency;
a matrix for each unique pair of items, wherein said matrix tallies ratings for each item of said pair;
means for computing a correlation between items of said pair from said matrix;
means for filtering non-significant correlations;
a list of correlating items, said list comprising a list of all significant correlations, wherein said list is periodically transmitted to at least one client from said server side; and
client-side means for predicting a rating for an unrated item based on the correlations provided in the list of correlating items. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40)
-
-
23. A method of predicting items likely to appeal to a user, based on a distributed collaborative filtering engine, wherein ratings are predicted for said items on the client side using correlation factors downloaded from the server, said correlation factors being computed on the server side from preference profiles anonymously posted to said server from a plurality of clients, said method comprising the steps of:
-
providing a server;
providing a plurality of clients, wherein said clients are intermittently in contact with said server;
providing a population of items, wherein said user has rated a portion of said population;
periodically transmitting a list of said rated items to said server side over a stateless network connection;
aggregating said list with lists from other clients into a single list;
filtering said list based on frequency;
providing a matrix for each unique pair of items, wherein said matrix tallies ratings for each item of said pair;
computing a correlation from said matrix between items of said pair;
filtering out non-significant correlations;
compiling a list of correlating items comprising a list of all non-significant correlations;
periodically transmitting said list to said clients;
on the client side, predicting a rating for an unrated item, based on the correlation provided in the list of correlating items.
-
Specification