User-specific media playlists
A media recommendation system may score media items according to user recommendations, popularity, and/or recency. The scores may be weighted to produce an overall score for each media item. Media items may be added to a pool for a specific user, from which media items are selected for playback. The contents of the pool may be modified based upon user feedback and other data. The pool may be modified dynamically and/or in real time as media items are consumed or rated by the user.
1. A computer-implemented method comprising:
selecting a plurality of seed media items from among a plurality of media items available for playback to a user within a computerized media recommendation system;
providing a representation of each seed media item of the plurality of seed media items to a user;
receiving a selection of a first seed media item from among the plurality of seed media items from the user;
automatically generating a pool of media items based upon the first seed media item, the pool comprising a subset, less than the whole, of the plurality of media items available within the computerized media recommendation system;
generating a vector representation of each media item in the pool;
sorting the pool based upon the distance between the vector representations and a vector representation of the first seed media item; and
subsequent to beginning playback of the first seed media item, playing the first media item in the pool according to the sorting.
2. The computer-implemented method of claim 1 , wherein each of the plurality of seed media items is of a type selected from the group consisting of: trending, emerging, recent, liked, selected, active, and recommended.
3. The computer-implemented method of claim 1 , wherein the representations of the plurality of seed media items are presented to the user in a single user interface.
4. The computer-implemented method of claim 1 , wherein the automatically generating the pool further comprises:
generating a matrix M of users and media items, wherein a matrix element at the uth row and ith column corresponds to the number of times a user u has consumed the media item i; and minimizing the sum
∑
i
(
M
u
,
s
-
v
→
u
·
v
→
i
)
2
wherein {right arrow over (v u )} is a vector representation of the user and {right arrow over (v t )} is a vector representation of the ith media item.
5. The computer-implemented method of claim 1 , wherein the plurality of seed media items comprises 10-20 media items.
6. The computer-implemented method of claim 1 , wherein at least one of the plurality of seed media items is
selected from a previous pool generated for the user.
7. The computer-implemented method of claim 1 , wherein at least one of the plurality of seed media items is based upon the time of day the user accesses the computerized media recommendation system.
8. The computer-implemented method of claim 1 , further comprising:
generating a plurality of clusters, each of the plurality of clusters including a plurality of media items available within the computerized media recommendation system, based upon artist vectors associated artists of each of the plurality of media items available within the computerized media recommendation system.
9. The computer-implemented method of claim 8 , wherein at least one media item in the pool is selected based upon a cluster within which an artist of the at least one media item is located.
10. The computer-implemented method of claim 1 , further comprising:
receiving a first feedback regarding the first seed media item from the user;
modifying the contents of the pool based upon the first feedback to generate a modified pool, wherein the modified pool includes at least one media item not included in the pool, excluded at least one media item included in the pool, or both.
11. A computerized media item recommendation system, comprising:
a media item processing system including a processor communicatively coupled to a memory that is configured to:
select a plurality of seed media items from among a plurality of media items available for playback to a user within a computerized media recommendation system;
provide a representation of each seed media item of the plurality of seed media items to a user;
receive a selection of a first seed media item from among the plurality of seed media items from the user;
automatically generate a pool of media items based upon the first seed media item, the pool comprising a subset, less than the whole, of the plurality of media items available within the computerized media recommendation system; and
a recommendation model training system configured to:
generate a vector representation of each media item in the pool; and
sort the pool based upon the distance between the vector representations and a vector representation of the first seed media item;
wherein the computerized media item recommendation system is configured to play the first seed media item and, subsequent to beginning playback of the first seed media item, playing the first media item in the pool according to the sorting.
12. The system of claim 11 , wherein each of the plurality of seed media items is of a type selected from the group consisting of: trending, emerging, recent, liked, selected, active, and recommended.
13. The system of claim 11 , wherein the representations of the plurality of seed media items are presented to the user in a single user interface.
14. The system of claim 11 , wherein the pool is generated by:
generating a matrix M of users and media items, wherein a matrix element at the uth row and ith column corresponds to the number of times a user u has consumed the media item i; and minimizing the sum
∑
i
(
M
u
,
s
-
v
→
u
·
v
→
i
)
2
wherein {right arrow over (v u )} is a vector representation of the user and {right arrow over (v t )} is a vector representation of the ith media item.
15. The system of claim 11 , wherein the plurality of seed media items comprises 10-20 media items.
16. The system of claim 11 , wherein at least one of the plurality of seed media items is selected from a previous pool generated for the user.
17. The system of claim 11 , wherein at least one of the plurality of seed media items is based upon the time of day the user accesses the computerized media recommendation system.
18. The system of claim 11 , further configured to:
generate a plurality of clusters, each of the plurality of clusters including a plurality of media items available within the computerized media recommendation system, based upon artist vectors associated artists of each of the plurality of media items available within the computerized media recommendation system.
19. The system of claim 18 , wherein at least one media item in the pool is selected based upon a cluster within which an artist of the at least one media item is located.
20. The system of claim 10 , further configured to:
receive a first feedback regarding the first seed media item from the user; and
modify the contents of the pool based upon the first feedback to generate a modified pool, wherein the modified pool includes at least one media item not included in the pool, excluded at least one media item included in the pool, or both.