Model framework for media content recommendations
Technologies are described herein for media content recommendations. For example, a method of media content recommendations may include generating, by a plurality of machine learning algorithms, a plurality of different sets of media content recommendations based on a content consumption history associated with a user account, generating a first playlist based on a first pair of the plurality of sets of media content recommendations, generating a second playlist based on a second pair of the plurality of sets of media content recommendations, wherein the first pair and the second pair are different from one another, and providing, via the API, the first playlist and the second playlist to at least one of the one or more media applications.
1 . A computer-implemented method, comprising:
generating, by a plurality of machine learning algorithms, a plurality of different sets of media content recommendations based on a content consumption history associated with a user account, wherein individual sets of media content recommendations include at least one different media content recommendation from other sets of the media content recommendations;
generating, in response to user input associated with one or more media applications and received from the one or more media applications via an application programming interface (API), a first playlist based on a first pair of the plurality of different sets of media content recommendations;
generating, in response to receipt of an identifier of a computing device associated with the user account, a second playlist based on a second pair of the plurality of different sets of media content recommendations, wherein the first pair and the second pair are different from one another; and
providing, via the API, the first playlist and the second playlist to at least one of the one or more media applications.
2 . The computer-implemented method of claim 1 , wherein the plurality of machine learning algorithms implement different machine learning models chosen from collaborative filtering models, two-tower models, and sequence models.
3 . The computer-implemented method of claim 1 , wherein the plurality of machine learning algorithms include one or more of: collaborative filtering models, two-tower models, or sequence models.
4 . The computer-implemented method of claim 1 , wherein the content consumption history associated with the user account comprises one or more of: media consumption history, genre consumption history, song consumption history, and playlist engagement history.
5 . The computer-implemented method of claim 1 , further comprising storing additional user inputs associated with any of the provided first playlist and the provided second playlist as additional engagement data.
6 . The computer-implemented method of claim 5 , wherein the additional engagement data comprises one or more of: length of playback of a particular playlist, length of playback of a particular media content item, repetitive plays of a particular media content item, continued user engagement with a particular playlist, and continued user engagement with a particular musical genre.
7 . The computer-implemented method of claim 5 , further comprising augmenting at least one of the first playlist and the second playlist based on at least the additional engagement data.
8 . The computer-implemented method of claim 1 , further comprising, prior to generating the first playlist and the second playlist, caching the plurality of different sets of media content recommendations.
9 . The computer-implemented method of claim 1 , wherein at least one of generating the first playlist or generating the second playlist is further based on user context.
10 . The computer-implemented method of claim 9 , wherein the user context comprises one or more of: intent of a listening session, mood of a current user, and activity indication of a user device.
11 . A system comprising:
one or more processors; and
one or more memories, coupled to the one or more processors and having computer-readable instructions stored thereon, which when executed by the one or more processors, cause the one or more processors to:
generate, by a plurality of machine learning algorithms, a plurality of different sets of media content recommendations based on a content consumption history associated with a user account, wherein individual sets of media content recommendations include at least one different media content recommendation from other sets of the media content recommendations;
generate, in response to user input associated with one or more media applications and received from the one or more media applications via an application programming interface (API), a first playlist based on a first pair of the plurality of different sets of media content recommendations;
generate, in response to receipt of an identifier of a computing device associated with the user account, a second playlist based on a second pair of the plurality of different sets of media content recommendations, wherein the first pair and the second pair are different from one another; and
provide, via the API, the first playlist and the second playlist to at least one of the one or more media applications.
12 . The system of claim 11 , wherein the plurality of machine learning algorithms implement different machine learning models chosen from collaborative filtering models, two-tower models, and sequence models.
13 . The system of claim 11 , wherein the plurality of machine learning algorithms include one or more of: collaborative filtering models, two-tower models, or sequence models.
14 . The system of claim 11 , wherein the content consumption history associated with the user account comprises one or more of: media consumption history, genre consumption history, song consumption history, and playlist engagement history.
15 . The system of claim 11 , wherein the computer-readable instructions further cause the system to store additional user inputs associated with any of the provided first playlist and the provided second playlist as additional engagement data.
16 . The system of claim 15 , wherein the additional engagement data comprises one or more of: length of playback of a particular playlist, length of playback of a particular media content item, repetitive plays of a particular media content item, continued user engagement with a particular playlist, and continued user engagement with a particular musical genre.
17 . The system of claim 15 , wherein the computer-readable instructions further cause the system to augment at least one of the first playlist and the second playlist based on at least the additional engagement data.
18 . The system of claim 11 , wherein the first playlist and the second playlist are further based on user context.
19 . The system of claim 18 , wherein the user context comprises one or more of: intent of a listening session, mood of a current user, and activity indication of a user device.
20 . A non-transitory computer-readable medium with instructions that, when executed by one or more processors at a device, cause the one or more processors to perform operations, the operations comprising:
generating, by a plurality of machine learning algorithms, a plurality of different sets of media content recommendations based on a content consumption history associated with a user account, wherein individual sets of media content recommendations include at least one different media content recommendation from other sets of the media content recommendations;
generating, in response to user input associated with one or more media applications and received from the one or more media applications via an application programming interface (API), a first playlist based on a first pair of the plurality of different sets of media content recommendations;
generating, in response to receipt of an identifier of a computing device associated with the user account, a second playlist based on a second pair of the plurality of different sets of media content recommendations, wherein the first pair and the second pair are different from one another; and
providing, via the API, the first playlist and the second playlist to at least one of the one or more media applications.