Dynamic feedback in a recommendation system
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 method comprising:
changing, by a processor, a profile vector of a user in a manner that is inversely proportional to a total number of times that the user has consumed a specific media item of a plurality of media items, wherein the profile vector is based upon vector representations of the plurality of media items previously consumed by the user;
selecting, by a processor and based on the profile vector, a seed media item from among the plurality of media items, wherein a selection of the seed media item is based on a selection of a type of seed media item received by the processor, and wherein the type of seed media item is at least one selected from the group consisting of: a trending media item, and an emerging media item;
generating, by the processor, a first pool of media items based upon the seed media item, the first pool including a second subset of the plurality of media items;
receiving, by the processor, a feedback signal about a media item of the first pool; and
modifying, by the processor and based on the feedback signal, contents of the first pool to produce a second pool, wherein the second pool includes a media item not included in the first pool.
2. The method of claim 1 , further comprising:
generating the profile vector of the user based upon the vector representations of the plurality media items previously consumed by the user.
3. The method of claim 1 , wherein the specific media item is one of the plurality of media items previously consumed by the user.
4. The method of claim 1 , wherein the type of seed media item is at least one selected from the group consisting of: a recent media item, a liked media item, a selected media item, an active media item, and a recommended media item.
5. The method of claim 1 , wherein the profile vector of the user comprises a normalized sum of media item vectors.
6. The method of claim 5 , wherein the vector representation is produced from at least one of a word embedding process performed on the media items previously consumed by the user, a neural network process performed on the media items previously consumed by the user, or a collaborative filtering process performed on the media items previously consumed by the user.
7. The method of claim 5 , further comprising producing the profile vector of the user in real time.
8. The method of claim 1 , wherein the plurality of media items are available for playback within a computerized media recommendation system.
9. The method of claim 8 , wherein the profile vector of the user is based on media items, of the computerized media recommendation system, previously consumed by the user.
10. The method of claim 1 , further comprising:
generating, by the processor, vectors for the first pool; and
sorting, by the processor and based on distances between a vector for the seed media item and the vectors for the first pool.
11. The method of claim 10 , wherein:
the modifying occurs after the generating the vectors for the first pool; and
the modifying occurs before the sorting.
12. The method of claim 1 , further comprising selecting, by the processor, a first subsequent media item.
13. The method of claim 12 , further comprising determining, by the processor and after a selection of the first subsequent media item, a lack of a receipt of a feedback signal for the first subsequent media item.
14. The method of claim 13 , further comprising selecting, by the processor and after a determination of the lack of the receipt of the feedback signal for the first subsequent media item, a second subsequent media item.
15. The method of claim 12 , further comprising determining, by the processor and after a selection of the first subsequent media item, a receipt of a feedback signal for the first subsequent media item.
16. The method of claim 15 , wherein the modifying occurs after a determination of the receipt of the feedback signal for the first subsequent media item.
17. The method of claim 1 , further comprising producing, by the processor, the profile vector of the user.
18. The method of claim 17 , wherein the producing comprises producing the profile vector of the user by a collaborative filtering process.