Recommendation engine apparatus and methods
Recommendation engine apparatus and associated methods provide content compiled from various sources and selected to match user preferences. In one embodiment, the recommendation apparatus comprises a headend entity; in another, it is co-located on a user's CPE. In one embodiment, the recommendation engine creates content records from content metadata for comparison to a user profile. The user profile is pre-programmed; however has the ability to dynamically shift toward a user's preferences as the user takes actions regarding content. Client applications are utilized to compile and present content; feedback mechanisms are utilized to enable “learning” from user activities to generate more precise recommendations as well as to “unlearn” stale preferences. Recommended content is displayed in the form of a playlist, or as a continuous stream on a virtual channel, or presented in an electronic program guide. A business rules “engine” useful in implementing operational or business goals is also disclosed.
1. For use in a content delivery network, a recommendation apparatus comprising:
a storage device adapted to:
store a plurality of information regarding a plurality of content elements expressed as n×1 column vectors, said n×1 column vectors each comprising a number, n, of aspects;
store a plurality of user profiles expressed as n×1 column vectors, said n×1 column vectors each comprising a number, n, of aspects; and
store an n×1 column reference vector configured to define each of said n aspects as expressed in said plurality of content element vectors and said plurality of user profile vectors, where n is more than one; and
a digital processor in data communication with said storage device and adapted to run a computer program thereon, said computer program adapted to:
compare said plurality of content element column vectors to individual ones of said plurality of user profile column vectors; and
based at least in part on said comparison, produce at least one list of a plurality of content elements, said each of said plurality of content elements bearing a relationship to at least one of said plurality of user profiles.
2. The recommendation apparatus of claim 1 , wherein said recommendation apparatus comprises a headend entity of said network.
3. The recommendation apparatus of claim 1 , wherein said recommendation apparatus comprises a hub entity of said network.
4. The recommendation apparatus of claim 1 , wherein said recommendation apparatus comprises a consumer premises device.
5. The recommendation apparatus of claim 1 , wherein said aspects each comprise one or more aspects corresponding to metadata associated with individual ones of said plurality of content elements.
6. The recommendation apparatus of claim 5 , wherein said comparison of said plurality of content element column vectors to said plurality of user profile column vectors comprises calculation of a dot product of individual ones of said content element column vectors and individual ones of said user profile column vectors to produce a scalar quantity.
7. The recommendation apparatus of claim 6 , wherein said at least one list of content comprises a list of content having said scalar quantity greater than or equal to a preset real-value number, and said list being provided to a client device associated with said at least one user profile.
8. The recommendation apparatus of claim 1 , wherein said list comprises information for identification and retrieval of each of said content in said list.
9. The recommendation apparatus of claim 1 , wherein at least one of said plurality of user profiles comprises a template user profile.
10. The recommendation apparatus of claim 5 , wherein said program is further adapted to update said at least one user profile via addition of a training vector to individual ones of said user profile column vectors, said training vector comprising weighted data regarding one or more user actions.
11. A method of identifying content targeted to a user based on one or more actions of said user, said method comprising:
maintaining a user profile expressed as a vector at a storage apparatus of a recommendation apparatus, said user profile being specific to said user;
generating at a processor of said recommendation apparatus, a first plurality of content records expressed as vectors, said first plurality of content record vectors being based, at least in part, on metadata regarding a plurality of content, each of said first plurality of content record vectors, and said user profile vector each having identical numbers of both columns and rows as one another;
providing access to said plurality of content to said user;
storing a user activity vector at said storage apparatus, said user activity vector comprising information relating to at least one action taken by said user with respect to an individual one of said plurality of content;
generating at said processor of said recommendation engine, a second plurality of content record vectors, individual ones of said second plurality of content record vectors each comprising a weighted content record of said content acted upon by said user expressed as a vector;
utilizing said individual ones of said second plurality of content record vectors at least in part to generate a second user profile vector;
calculating a product of said individual ones of said second plurality of content record vectors to said second user profile vector to produce a scalar quantity; and
said recommendation engine identifying individual ones of said plurality of content having a prescribed relation to an updated user profile based at least in part on said scalar quantity.
12. The method of claim 11 , wherein said user profile comprises a pre-set or template user profile.
13. The method of claim 11 , wherein said plurality of content record vectors and said user profile vector are expressed as n×1 column vectors; and said method further comprises storing a n×1 column reference vector defining aspects expressed in said plurality of content record vectors and said user profile vector.
14. The method of claim 11 , wherein said act of generating said second plurality of content record vectors comprises multiplying said content record vectors associated with said at least one act by a weighting factor.
15. The method of claim 11 , wherein said act of identifying individual ones of said plurality of content having a prescribed relation to said updated user profile based at least in part on said scalar quantity comprises identifying individual ones of said second plurality of content vectors having a scalar quantity equal to or greater than a predetermined value.
16. The method of claim 11 , further comprising displaying identifying information regarding said identified individual ones of said plurality of content having a prescribed relation to said updated user profile.
17. Computer readable apparatus comprising media adapted to contain a computer program having a plurality of instructions, said plurality of instructions which, when executed:
maintain at least one user profile;
generate a plurality of content records via at least metadata relating to a respective plurality of content;
provide access to said plurality of content to a user;
store a record of at least one act taken by said user with respect to an individual one of said plurality of content;
multiply an individual one of said plurality of content records associated with said individual one of said plurality of content by a weighting factor determined by a nature of said act and an estimated relation of said act to one or more user preferences to generate a second content record; and
add said second content record to said user profile to generate a second user profile.
18. The computer readable apparatus of claim 17 , wherein said user profile comprises a pre-set or template user profile.
19. The computer readable apparatus of claim 17 , wherein said plurality of content records and said user profile are expressed as vectors having identical numbers of columns and rows as one another and said plurality of instructions are further configured to, when executed, compare first attributes of said second user profile to substantially equivalent first attributes of individual ones of said plurality of content records to identify individual ones thereof having a prescribed level of relatedness thereto.
20. The computer readable apparatus of claim 19 , wherein said comparison of said second user profile to individual ones of said content records comprises calculation of a dot product of individual ones of said content records to said user profile to produce a scalar quantity.
21. The computer readable apparatus of claim 20 , wherein said identification of said individual ones of said plurality of content having a prescribed level of relatedness comprises identification of individual ones of said plurality of content having a scalar quantity equal to or greater than a predetermined value.
22. Computer readable apparatus comprising media adapted to contain a computer program having a plurality of instructions, said plurality of instructions which, when executed:
maintain a user profile expressed as a vector at a storage apparatus of a recommendation apparatus, said user profile configured to be specific to a user;
generate at a processor of said recommendation apparatus, a first plurality of content records expressed as vectors, said first plurality of content record vectors being based, at least in part, on metadata which relates to a plurality of content, each of said first plurality of content record vectors, and said user profile vector each comprise identical numbers of both columns and rows as one another;
provide access to said plurality of content to said user;
store a user activity vector at said storage apparatus, said user activity vector comprises information which relates to at least one action taken by said user with respect to an individual one of said plurality of content;
generate at said processor of said recommendation engine, a second plurality of content record vectors, individual ones of said second plurality of content record vectors each comprise a weighted content record of said content acted upon by said user expressed as a vector;
utilize said individual ones of said second plurality of content record vectors at least in part to generate a second user profile vector;
calculate a product of said individual ones of said second plurality of content record vectors to said second user profile vector to produce a scalar quantity; and
identify individual ones of said plurality of content which comprises a prescribed relation to an updated user profile based at least in part on said scalar quantity.
23. The computer readable apparatus of claim 22 , wherein said user profile comprises a pre-set or template user profile.
24. The computer readable apparatus of claim 22 , wherein said plurality of content record vectors and said user profile vector are expressed as n×1 column vectors; and said plurality of instructions are further configured to, when executed, store a n×1 column reference vector which defines aspects expressed in said plurality of content record vectors and said user profile vector.
25. The computer readable apparatus of claim 22 , wherein said generation of said second plurality of content record vectors comprises multiplication of said content record vectors associated with said at least one act by a weighting factor.
26. The computer readable apparatus of claim 22 , wherein said identification of individual ones of said plurality of content comprise a prescribed relation to said updated user profile based at least in part on said scalar quantity comprises identification of individual ones of said second plurality of content vectors which comprise a scalar quantity equal to or greater than a predetermined value.
27. The computer readable apparatus of claim 22 , wherein said plurality of instructions are further configured to, when executed, display identification information which relates to said identified individual ones of said plurality of content which comprise a prescribed relation to said updated user profile.