Optimal greedy diversity for recommendation

The need for diversification manifests in various recommendation use cases. In this work, we pro-pose a novel approach to diversifying a list of rec-ommended items, which maximizes the utility of the items subject to the increase in their diversity. From a technical perspective, the problem can be viewed as maximization of a modular function on the polytope of a submodular function, which can be solved optimally by a greedy method. We eval-uate our approach in an offline analysis, which in-corporates a number of baselines and metrics, and in two online user studies. In all the experiments, our method outperforms the baseline methods. 1