Towards Persuasive Recommender Systems

Abstract

The primary objective of recommender systems, in a general sense, is to recommend items to users rather than to persuade users to get those items. Hence, a recommender system is not a persuasive technology by itself. In recent years, however, there has been increasing attention in the literature towards augmenting persuasiveness features into recommender systems. Several researchers have discussed the feasibility of enriching recommendations with persuasive messages. Nonetheless, there is a lack of works that discuss how to incorporate personalized and dynamic persuasive capabilities to recommender systems. To mitigate this issue, we propose the Personalized Persuasive RS (PerPer) framework. The PerPer adopts learning automata concepts to dynamically choose a suitable persuasive strategy for users in a personalized manner. PerPer is general enough to be plugged into different recommenders and to consider several persuasive strategies. PerPer aims to provide a simple and straightforward way to incorporate persuasive features to recommenders. By this, it would have the potential of increasing users’ perceived acceptance of recommendations