Media channel creation based on free-form media input seeds

1. A computer-implemented method for generating a media channel including a plurality of media items, comprising:

storing a set of media items, each media item stored with a predetermined plurality of media item scores, each media item score for the media item describing a magnitude of the media item's relevance to a distinct ambiguous entity term from a set of ambiguous entity terms;

storing a plurality of predetermined entity scores, each entity score associated with a corresponding ambiguous entity term from the set of ambiguous entity terms and describing an importance of the corresponding ambiguous entity term in media input seeds;

receiving a request for a media channel from a client device of a user, the request including a textual free-form media input seed that comprises a plurality of ambiguous entity terms input by the user, wherein an ambiguous entity term of the plurality of ambiguous entity terms does not identify an entity within a set of valid entity types related to the set of media items, the valid set of entity types having meaning within a music context and the valid set of entity types including a musicological feature;

identifying, for each ambiguous entity term from the plurality of ambiguous entity terms input by the user, a plurality of media items that are correlated with the ambiguous entity term from the set of media items, the correlation of a media item to an ambiguous entity term based on the media item's predetermined media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

generating, for each ambiguous entity term, a media playlist to generate a plurality of media playlists, each media playlist including the identified plurality of media items that are correlated with the ambiguous entity term;

calculating ranking scores for the identified plurality of media items included in the plurality of media playlists, a ranking score for a media item from the identified plurality of media items based on a stored entity score for the ambiguous entity term that is associated with the media playlist that includes the media item and the media item's media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

ranking the identified plurality of media items based on the calculated ranking scores; and

combining the plurality of media playlists for the plurality of ambiguous entity terms included in the textual free-form media input seed into the media channel by selecting a subset of the identified plurality of media items to include in the media channel based on the ranking; and

providing the media channel to the client device.

2. The computer-implemented method of claim 1 , wherein generating the media playlist further comprises:

identifying musicological features of a media item in the media playlist;

identifying other media items having musicological features matching the musicological features of the media item in the media playlist; and

adding the identified other media items to the media playlist.

3. The computer-implemented method of claim 1 , wherein calculating the ranking scores comprises:

identifying a co-occurring media item that is included in multiple ones of the plurality of media playlist;

calculating a ranking score for the co-occurring media item based on entity scores respectively describing importances of the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included and media item scores indicating magnitudes of the co-occurring media item's relevance to the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included.

4. The computer-implemented method of claim of claim 1 , further comprising:

receiving user feedback on the subset of the identified plurality of media items provided in the media channel;

adjusting entity scores of the plurality of ambiguous entity terms and media item scores of the subset of the identified plurality of media items according to the user feedback;

adjusting the ranking scores for the subset of the identified plurality of media items based on the adjusted entity scores and the adjusted media item scores;

updating the ranking of the subset of the identified plurality of media items based on the adjusted ranking scores; and

selecting another subset of the identified plurality of media items to include in the media channel based on the updated ranking.

5. A non-transitory computer-readable storage medium storing executable computer program instructions for generating a media channel including a plurality of media items, the instructions executable to perform steps comprising:

storing a set of media items, each media item stored with a predetermined plurality of media item scores, each media item score for the media item describing a magnitude of the media item's relevance to a distinct ambiguous entity term from a set of ambiguous entity terms;

storing a plurality of predetermined entity scores, each entity score associated with a corresponding ambiguous entity term from the set of ambiguous entity terms and describing an importance of the corresponding ambiguous entity term in media input seeds;

receiving a request for a media channel from a client device of a user, the request including a textual free-form media input seed that comprises a plurality of ambiguous entity terms input by the user, wherein an ambiguous entity term of the plurality of ambiguous entity terms does not identify an entity within a set of valid entity types related to the set of media items, the valid set of entity types having meaning within a music context and the valid set of entity types including a musicological feature;

identifying, for each ambiguous entity term from the plurality of ambiguous entity terms input by the user, a plurality of media items that are correlated with the ambiguous entity term from the set of media items, the correlation of a media item to an ambiguous entity term based on the media item's predetermined media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

generating, for each ambiguous entity term, a media playlist to generate a plurality of media playlists, each media play list including the identified plurality of media items that are correlated with the ambiguous entity term;

calculating ranking scores for the identified plurality of media items included in the plurality of media playlists, a ranking score for a media item from the identified plurality of media items based on a stored entity score for the ambiguous entity term that is associated with the media playlist that includes the media item and the media item's media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

ranking the identified plurality of media items based on the calculated ranking scores; and

combining the plurality of media playlists for the plurality of ambiguous entity terms included in the textual free-form media input seed into the media channel by selecting a subset of the identified plurality of media items to include in the media channel based on the ranking; and

providing the media channel to the client device.

6. The non-transitory computer-readable storage medium of claim 5 , wherein generating the media playlist further comprises:

identifying musicological features of a media item in the media playlist;

identifying other media items having musicological features matching the musicological features of the media item in the media playlist; and

adding the identified other media items to the media playlist.

7. The non-transitory computer-readable storage medium of claim 5 , wherein calculating the ranking scores comprises:

identifying a co-occurring media item that is included in multiple ones of the plurality of media playlist; and

calculating a ranking score for the co-occurring media item based on entity scores respectively describing importances of the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included and media item scores indicating magnitudes of the co-occurring media item's relevance to the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included.

8. The non-transitory computer-readable storage medium of claim 5 , wherein the instructions are executable to perform further steps comprising:

receiving user feedback on the subset of the identified plurality of media items provided in the media channel;

adjusting entity scores of the plurality of ambiguous entity terms and media item scores of the subset of the identified plurality of media items according to the user feedback;

adjusting the ranking scores for the subset of the identified plurality of media items based on the adjusted entity scores and the adjusted media item scores;

updating the ranking of the subset of the identified plurality of media items based on the adjusted ranking scores; and

selecting another subset of the identified plurality of media items to include in the media channel based on the updated ranking.

9. A device for generating a media channel including a plurality of media items, comprising:

a processor configured to execute modules; and

a memory storing the modules, the modules executable by the processor to perform steps comprising:

storing a set of media items, each media item stored with a predetermined plurality of media item scores, each media item score for the media item describing a magnitude of the media item's relevance to a distinct ambiguous entity term from a set of ambiguous entity terms;

storing a plurality of predetermined entity scores, each entity score associated with a corresponding ambiguous entity term from the set of ambiguous entity terms and describing an importance of the corresponding ambiguous entity term in media input seeds;

receiving a request for a media channel from a client device of a user, the request including a textual free-form media input seed that comprises a plurality of ambiguous entity terms input by the user, wherein an ambiguous entity term of the plurality of ambiguous entity terms does not identify an entity within a set of valid entity types related to the set of media items, the valid set of entity types having meaning within a music context and the valid set of entity types including a musicological feature;

identifying, for each ambiguous entity term from the plurality of ambiguous entity terms input by the user, a plurality of media items that are correlated with the ambiguous entity term from the set of media items, the correlation of a media item to an ambiguous entity term based on the media item's media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

generating, for each ambiguous entity term, a media playlist to generate a plurality of media playlists, each media play list including the identified plurality of media items that are correlated with the ambiguous entity term;

calculating ranking scores for the identified plurality of media items included in the plurality of media playlists, a ranking score for a media item from the identified plurality of media items based on a stored entity score for the ambiguous entity term that is associated with the media playlist that includes the media item and the media item's media item score indicating a magnitude of the media item's relevance to the ambiguous entity term;

ranking the identified plurality of media items based on the calculated ranking scores; and

combining the plurality of media playlists for the plurality of ambiguous entity terms included in the textual free-form media input seed into the media channel by selecting a subset of the identified plurality of media items to include in the media channel based on the ranking; and

providing the media channel to the client device.

10. The device of claim of claim 9 , wherein generating the playlist further comprises:

identifying musicological features of a media item in the media playlist;

identifying other media items having musicological features matching the musicological features of the media item in the media playlist; and

adding the identified other media items to the media playlist.

11. The device of claim 10 , wherein calculating the ranking scores comprises:

identifying a co-occurring media item that is included in multiple ones of the plurality of media playlist; and

calculating a ranking score for the co-occurring media item based on entity scores respectively describing importances of the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included and media item scores indicating magnitudes of the co-occurring media item's relevance to the ambiguous entities associated with the multiple ones of the plurality of media playlists in which the co-occurring media item is included.

12. The computer-implemented method of claim 1 , further comprising:

identifying, for at least one ambiguous entity term, a media item that includes the ambiguous entity term in lyrics of the media item.

13. The computer-implemented method of claim 1 , further comprising:

identifying, for at least one ambiguous entity term, a media item that includes the ambiguous entity term in a title of the media item.

14. The computer-implemented method of claim 1 , wherein identifying, for each ambiguous entity term a plurality of media items that are correlated with the ambiguous entity term comprises:

identifying for each ambiguous entity term, media items from the stored set of media items having media item scores for the ambiguous entity term that are above a threshold.

15. The computer-implemented method of claim 1 , wherein a first media item from the identified plurality of media items is associated with a first ambiguous entity term input by the user that has a higher entity score than a second ambiguous entity term input by the user that is associated with a second media item from the identified plurality of media items, and

wherein the first media item is ranked higher in the media channel than the second media item.