Enhancing Probability-based Recommender Systems with Recklessness Regularization
Incorporating a recklessness regularization term in the learning process of probability-based recommender systems enables controlling the risk level of predictions, leading to improved quantity and quality of recommendations.