Media recommendation using internet media stream modeling

Abstract

Media item recommendations, such as music track recommendations, may be made using one or more models generated using data collected from a plurality of media stream sources, such as, for example, Internet radio stations. In an initial, bootstrapping phase, data about media items and media stream playlists of media stream sources may be used to generate a model, which comprises latent factor vectors, or learned profiles, of media items, e.g., tracks, artists, etc. Such a bootstrapping phase may be performed without user data, such as user playlists and/or user feedback, to generate a model that may be used to make media item recommendations. As user data becomes available, e.g., as users of a recommendation service provide user data, the user data may be used to supplement and/or update the model and/or to create user profiles.

1 - 30 . (canceled) 
     
     
         31 . A method comprising:
 collecting, via at least one computing device, a training data set comprising data about a plurality of media items and a plurality of media streams of a number of Internet streaming media sources, for each occurrence of a media item of the plurality in a media stream of the plurality, the training data set comprising information about the media item and the media stream;   generating, via the at least one computing device, a latent factor model using the training data set, the latent factor model modeling the plurality of media items and the plurality of media streams, the latent factor model is at least initially generated without using user data identifying user media item preferences;   identifying, via the at least one computing device, a seed item; and   using, via the at least one computing device, the identified seed item and the latent factor model to make a number of item recommendations, the number of item recommendations comprising at least one item recommendation that is made using the latent factor model generated without using the user data.   
     
     
         32 . The method of  claim 31 , further comprising:
 receiving, via the at least one computing device and from a client computing device, input indicative of a user selection of the seed item.   
     
     
         33 . The method of  claim 31 , further comprising:
 randomly identifying, via the at least one computing device, the seed item.   
     
     
         34 . The method of  claim 31 , further comprising:
 identifying, via the at least one computing device, the seed item based on information about the user.   
     
     
         35 . The method of  claim 34 , the information about the user comprising user content preferences. 
     
     
         36 . The method of  claim 34 , the information about the user comprising at least one of age and gender. 
     
     
         37 . The method of  claim 31 , the using the latent factor model to make a recommendation, further comprising:
 comparing, via the at least one computing device, a latent factor vector, of the latent factor model, corresponding to the seed item with a number of latent factor vectors, of the latent factor model, to identify a number of items for inclusion in an item recommendation of the number of item recommendations, each latent factor vector of the number corresponding to an item being considered for inclusion in the item recommendation of the number of item recommendations.   
     
     
         38 . The method of  claim 37 , each item selected for inclusion in the item recommendation of the number of item recommendations is determined, using the latent factor model, to be dissimilar to the identified seed item. 
     
     
         39 . The method of  claim 37 , each item selected for inclusion in the item recommendation of the number of item recommendations is determined, using the latent factor model, to be similar to the identified seed item. 
     
     
         40 . The method of  claim 31 , the seed item is a media item of the plurality of media items and each item recommendation of the number of item recommendations comprising at least one of a number of media items of the plurality of media items and a number of media streams of the plurality of media streams. 
     
     
         41 . The method of  claim 31 , the seed item is a media stream of the plurality of media streams and each item recommendation of the number of item recommendations comprising at least one a number of media items of the plurality of media items and a number of media streams of the plurality of media streams. 
     
     
         42 . The method of  claim 31 , further comprising:
 obtaining, via the at least one computing device, user data identifying user media item preferences;   supplementing, via the at least one computing device and using the obtained user data identifying user media item preferences, the latent factor model initially generated without using the user data; and   using, via the at least one computing device, the identified seed item and the supplemented latent factor model, initially generated without using the user data and subsequently supplemented using the user data identifying user media item preferences, to make an item recommendation of the number of item recommendations.   
     
     
         43 . A non-transitory computer-readable storage medium tangibly encoded with computer-executable instructions that when executed by a processor associated with a computing device perform a method comprising:
 collecting a training data set comprising data about a plurality of media items and a plurality of media streams of a number of Internet streaming media sources, for each occurrence of a media item of the plurality in a media stream of the plurality, the training data set comprising information about the media item and the media stream;   generating a latent factor model using the training data set, the latent factor model modeling the plurality of media items and the plurality of media streams, the latent factor model is at least initially generated without using user data identifying user media item preferences;   identifying a seed item; and   using the identified seed item and the latent factor model to make a number of item recommendations, the number of item recommendations comprising at least one item recommendation that is made using the latent factor model generated without using the user data.   
     
     
         44 . The non-transitory computer-readable storage medium of  claim 43 , the using the latent factor model to make a recommendation, further comprising:
 comparing a latent factor vector, of the latent factor model, corresponding to the seed item with a number of latent factor vectors, of the latent factor model, to identify a number of items for inclusion in an item recommendation of the number, each latent factor vector of the number corresponding to an item being considered for inclusion in the item recommendation of the number of item recommendations.   
     
     
         45 . The non-transitory computer-readable storage medium of  claim 44 , each item selected for inclusion in the item recommendation of the number of item recommendations is determined, using the latent factor model, to be dissimilar to the identified seed item. 
     
     
         46 . The non-transitory computer-readable storage medium of  claim 44 , each item selected for inclusion in the item recommendation of the number of item recommendations is determined, using the latent factor model, to be similar to the identified seed item. 
     
     
         47 . The non-transitory computer-readable storage medium of  claim 43 , the seed item is a media item of the plurality of media items and each item recommendation of the number of item recommendations comprising at least one of a number of media items of the plurality of media items and a number of media streams of the plurality of media streams. 
     
     
         48 . The non-transitory computer-readable storage medium of  claim 43 , the seed item is a media stream of the plurality of media streams and each item recommendation of the number of item recommendations comprising at least one a number of media items of the plurality of media items and a number of media streams of the plurality of media streams. 
     
     
         49 . The non-transitory computer-readable storage medium of  claim 43 , further comprising:
 obtaining, via the at least one computing device, user data identifying user media item preferences;   supplementing, via the at least one computing device and using the obtained user data identifying user media item preferences, the latent factor model initially generated without using the user data; and   using, via the at least one computing device, the identified seed item and the supplemented latent factor model, initially generated without using the user data and subsequently supplemented using the user data identifying user media item preferences, to make an item recommendation of the number of item recommendations.   
     
     
         50 . A computing device comprising:
 a processor;   a non-transitory storage medium for tangibly storing thereon program logic for execution by the processor, the program logic comprising:   collecting logic executed by the processor for collecting a training data set comprising data about a plurality of media items and a plurality of media streams of a number of Internet streaming media sources, for each occurrence of a media item of the plurality in a media stream of the plurality, the training data set comprising information about the media item and the media stream;   generating logic executed by the processor for generating a latent factor model using the training data set, the latent factor model modeling the plurality of media items and the plurality of media streams, the latent factor model is at least initially generated without using user data identifying user media item preferences;   identifying logic executed by the processor for identifying a seed item; and   using logic executed by the processor for using the identified seed item and the latent factor model to make a number of item recommendations, the number of item recommendations comprising at least one item recommendation that is made using the latent factor model generated without using the user data.