A survey of diversification techniques in Recommendation Systems

Abstract

Recommendation Systems provide suggestions for items that are useful to a user. Initially researches in RS mainly focused to improve only accuracy of the system, however improving only accuracy does not improve user satisfaction. Recently, it has been identified that diversity is an important dimension for evaluating a recommendation system. Users find a diversified set of recommendations more interesting than a monotonous only relevance based recommendations. This paper focuses only on the diversification techniques introduced in recommendation systems. We studied papers and articles published in academic literature and categorized them into different categories. We have also highlighted trending directions that are being used to diversify recommendations.