Relational Boosted Bandits

Contextual bandits algorithms have become essential in real-world user interaction problems recent years. However, these represent context as attribute value representation, which makes them infeasible for real world domains like social networks, are inherently relational. We propose Relational Boosted Bandits (RB2), a contextual algorithm relational based on (relational) boosted trees. RB2 enables us to learn interpretable and explainable models due the more descriptive nature of representation. empirically demonstrate effectiveness interpretability tasks such link prediction, classification, recommendation.