Spotify Ab

Benjamin Lacker

Rishabh Mehrotra

Samantha Hansen

Hugues Bouchard

James E McInerney

Aloïs Gruson

Ian Anderson

Oguz Semerci

Margreth Mpossi

Clay Gibson

Scott Wolf

Santiago Gil

Will Shapiro

Methods, systems and computer program products are provided personalizing recommendations of items with associated explanations. The example embodiments described herein use contextual bandits to personalize explainable recommendations ('recsplanations') as treatments (“Bart”). Bart learns and predicts satisfaction (e.g., click-through rate, consumption probability) for any combination of item, explanation, and context and, through logging and contextual bandit retraining, can learn from its mistakes in an online setting.

Personalizing explainable recommendations with bandits

Methods, systems and computer program products are provided personalizing recommendations of items with associated explanations. The example embodiments described herein use contextual bandits to personalize explainable recommendations ('recsplanations') as treatments ('Bart'). Bart learns and predicts satisfaction (e.g., click-through rate, consumption probability) for any combination of item, explanation, and context and, through logging and contextual bandit retraining, can learn from its mistakes in an online setting.

An electronic device associated with a media-providing service has one or more processors and memory storing instructions for execution by the one or more processors. For each track of a plurality of tracks consumed by a user of the media-providing service, the electronic device receives, over a time window, an indication of user selection of the track to compile a listening history of the user and determines, based on the listening history of the user with the media-providing service, whether the track has previously been consumed by the user. The electronic device determines, for the user, a discovery score corresponding to an average preference of the user for selecting previously unconsumed tracks. In response to determining that an amount of use of the media-providing service by the user does not satisfy a use threshold, the electronic device forgoes recommending tracks to the user based on the discovery score.

Methods and systems for personalizing user experience based on use of service

An electronic device is associated with a media-providing service. The electronic has one or more processors and memory storing instructions for execution by the one or more processors. For each track of a plurality of tracks consumed by a user of the media-providing service, the electronic device determines, based on a listening history of the user with the media-providing service, whether the track has previously been consumed by the user. The electronic device determines, for the user, a discovery score based on a number of the plurality of tracks determined to not have been previously consumed by the user. The electronic device provides personalized content to the user based on the discovery score.

Methods and systems for personalizing user experience based on discovery metrics

Growing community of inventors

Brooklyn, NY, United States of America

Benjamin Lacker