Incorporating Aggregate Diversity in Recommender Systems Using Scalable Optimization Approaches

Abstract

The success of a recommender system is generally evaluated with respect to the accuracy of recommendations. However, recently diversity of recommendations has also become an important aspect in evaluating recommender systems. One dimension of diversity is called aggregate diversity which refers to the diversity of items in the recommendation lists of all users and can be defined with different metrics. The maximization of both accuracy and the aggregate diversity simultaneously renders a multi-objective optimization problem which can be handled by different approaches. In this paper, after providing a thorough analysis of the multi-objective optimization approaches for this problem, we propose a new model which takes into account both accuracy and aggregate diversity. Different from previous works our model is specifically designed to incorporate distributional diversity metrics, which measure how evenly the items are distributed in the recommendation lists of users. In order to solve the large-scale instances, we propose a column generation algorithm and a Lagrangian relaxation approach based on the decomposition of the model. We present the results of the mathematical models and the performance of the proposed methodology that are obtained by computational experiments on real world data sets. These results reveal that our model successfully captures the trade-off between the objectives and reaches very high levels of distributional diversity.

An adjustable re-ranking approach for improving the individual and aggregate diversities of product recommendations

“…Diversity_in_top_N) [19,20] or the distribution of recommended products among all recommendation lists (e.g. Gini_diversity) [14,15]. Diversity_ in_top_N favors recommending more products, but it does not consider the distribution of recommended products.…”

“…Such methods predict customers' preferences on products using previous accuracy oriented methods and then select products based on the predicted preferences and diversity oriented criteria. For instance, several studies [14,21,31] formulated the accuracy-diversity trade-off problem as a multi-objective optimization problem with certain constraints and proposed corresponding solutions to solve the optimization problem. It is possible for optimization methods to adjust the tradeoffs between recommendation accuracy and diversities by setting certain parameters, but it is very difficult to find the optimal solution [32,33].…”

“…Increasing AD, therefore having a higher possibility to recommend more products in the long tail, has great potential for gaining higher profits since products in the long tail are extremely abundant. From this point of view, AD is significant to ecommerce business [14,15].…”

Topic-based hierarchical Bayesian linear regression models for niche items recommendation

“…Accuracy-based recommendation algorithms, that is, collaborative filtering recommendation algorithms [4, 5] and matrix factorisation techniques [6], often suggest items which are popular among users and ignore items which are less popular. Recent studies have focused on other evaluation metrics including diversity [7, 8] and serendipity [9, 10] to evaluate how good the recommendations are [11]. It is important to dig out niche preferences of users in order to achieve high level of user satisfaction.…”

An integrated recommender system for improved accuracy and aggregate diversity

“…As unpopular items cost less to produce and have a higher profit margin (e.g., lower license fees of unpopular movies), they help to improve the aggregate diversity for the recommender system. Muter and Aytekin (2017) have introduced a scalable optimization approach for improving aggregate diversity. However, in most of the studies, accuracy has been decreasing with increasing aggregate diversity.…”