Personalizing explainable recommendations with bandits

US11301513B2 · kind B2 · utility

Assignee

• Spotify AB · SE

Inventors

• James E. McInerney · Jersey City, US
• Benjamin Lacker · Brooklyn, US
• Samantha Hansen · New York, US
• Aloïs Gruson · New York, US
• Rishabh Mehrotra · London, GB
• Hugues Bouchard · New York, US

Key dates

• Filing date: Jul 3, 2019
• Grant date: Apr 12, 2022
• Expiry date: Mar 22, 2040

Classification

• Technology area (CPC G): Physics
• CPC primary: G06N3/08
• WIPO field: Computer technology
• WIPO sector: Electrical engineering

Abstract

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.