Steering for unstructured media stations

Patent metadata

Field	Value
Publication number	US-11651204-B2
Application number	US-201815885047-A
Country	US
Kind code	B2
Filing date	Jan 31, 2018
Priority date	Sep 9, 2017
Publication date	May 16, 2023
Grant date	May 16, 2023

Abstract

The present technology pertains to steering a playlisting service toward media items that are likely to receive positive feedback from a user operating a client device. The present technology permits a request to play media items without requiring an input context. A playlist service can begin to receive feedback on the playback of the media items and the received playback can be utilized by a steering service in response to a steering request to identify media items for playback that are likely to receive positive feedback based on the feedback received on a sequence of previously played media items.

First claim

The invention claimed is:

1. At least one non-transitory computer readable medium comprising instructions stored on the computer readable medium that when executed cause a computing system to:
receive a request to play unspecified media items from a client device;
transmit a first collection of media items for playback to the client device;
receive, during playback of the first collection of media items, a steering request and feedback from the client device, the feedback including a sequence of playback of the first collection of media items and a feedback sequence for the first collection of media items, wherein the feedback sequence indicates a skipped status or a played status for each of a plurality of media items of the collection of media items in the sequence of playback, and
wherein the steering request indicates a request to modify a criteria by which media items are selected for playback; and
based on the steering request and the feedback sequence, determine a second media item as a candidate media item for playback by the client device, wherein the second media item was not included with the first collection of media items, and wherein the second media item is determined based on an ordering of the feedback sequence and is further based on a score of how likely the second media item will receive positive feedback after playback of the first collection of media items.

2. The at least one non-transitory computer readable medium of claim 1 , wherein the determination of the second media item as the candidate media item for playback by the client device is based on representations and the feedback sequence and wherein the representations describe characteristics of the plurality of media items for which feedback was received.

3. The at least one non-transitory computer readable medium of claim 1 , wherein the determination of the second media item is based on user profile information, the user profile information identifying one or more media item preferences of a user that provided the request to play unspecified media items.

4. The at least one non-transitory computer readable medium of claim 1 , wherein the feedback includes positive feedback regarding playback of at least one media item from the first collection of media items in addition to negative feedback, and the sequence of playback of the first collection of media items.

5. The at least one non-transitory computer readable medium of claim 1 , wherein the determination of the second media item is performed using a machine learned function.

6. The at least one non-transitory computer readable medium of claim 5 , wherein the instructions are effective to cause the computing system to train a machine learning system to yield the machine learned function, the instructions cause the computing system to:
input feedback data for media items played in a historical sequence to the machine learning system, where some of the feedback data for the media items played in the historical sequence is negative feedback;
learn, using the machine learning system, which media items were played in the historical sequence after the negative feedback; and
output the machine learned function by the machine learning system.

7. A method comprising:
sending first media items for playback to a client device;
recording playback data for the first media items, the playback data includes a feedback sequence for the first media items, wherein the feedback sequence indicates a skipped status or a played status for each of a plurality of media items of the first media items; and
after receiving a steering request, identifying second media items as candidates for playback based on an analysis of the recorded playback data, wherein the steering request indicates a request to modify a criteria by which media items are selected for playback, and wherein the second media item is determined based on an ordering of the feedback sequence and is further based on a score of how likely the second media item will receive positive feedback after playback of the first collection of media items.

8. The method of claim 7 comprising:
receiving input feedback data for media items played in a historical sequence, where some of feedback data for the media items played in the historical sequence is negative feedback, and
learning which media items were played in the historical sequence after the negative feedback, and
outputting a machine learned function for use in the analysis of the recorded playback data to identify the second media items as the candidates.

9. The method of claim 8 , wherein the input feedback data also includes user profile information for a user that provided the feedback data for the media items played in the historical sequence, the user profile information identifying one or more media item preferences of the user.

10. The method of claim 8 comprising:
inputting representations to the machine learned function, the representations describing characteristics of the first media items for which feedback data was received for the media items played in the historical sequence.

11. The method of claim 8 , wherein the media items played in the historical sequence are located in an audio characteristics embedding space, the method comprising:
inputting the feedback data for media items played in the historical sequence into the machine learned function along with an audio characteristic representation from the audio characteristics embedding space, whereby the machine learned function can score the second media items as candidates for playback with audio characteristics that are likely to result in a completed playback.

12. The method of claim 8 , wherein the media items played in the historical sequence are located in a behavioral characteristics embedding space, the method comprising:
inputting the feedback data for media items played in the historical sequence into the machine learned function along with a behavioral characteristic representation from the behavioral characteristics embedding space, whereby the machine learned function can score the second media items as candidates for playback with behavioral characteristics that are likely to result in a completed playback.

13. A system comprising:
at least one processor; and at least one storage comprising instructions stored on the storage that when executed cause the system to:
receive a request to play unspecified media items from a client device;
transmit a first collection of media items for playback to the client device;
receive, during playback of the first collection of media items, a steering request and feedback from the client device, the feedback including a sequence of playback of the first collection of media items and a feedback sequence for the first collection of media items, wherein the feedback sequence indicates a skipped status or a played status for each of a plurality of media items of the collection of media items in the sequence of playback, and
wherein the steering request indicates a request to modify a criteria by which media items are selected for playback; and
based on the steering request and the feedback sequence, determine a second media item as a candidate media item for playback by the client device, wherein the second media item was not included with the first collection of media items, and wherein the second media item is determined based on an ordering of the feedback sequence and is further based on a score of how likely the second media item will receive positive feedback after playback of the first collection of media items.

14. The system of claim 13 , wherein the determination of the se

Assignees

• Apple Inc

Inventors

• Levy Mark H
• Van Balen Jan Markus Harrie
• Shenoy Arvind S
• Mauch Matthias
• Cartoon Daniel

Classifications

• G06V10/82 Primary

  using neural networks · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06F16/4387 Primary

  by the use of playlists · CPC title

• G06F18/24143

  Distances to neighbourhood prototypes, e.g. restricted Coulomb energy networks [RCEN] · CPC title