Fuzzy Group Decision Making for influence-aware recommendations.

Group Recommender Systems are special kinds of Recommender Systems aimed at suggesting items to groups rather than individuals taking into account, at the same time, the preferences of all (or the majority of) members. Most existing models build recommendations for a group by aggregating the preferences for their members without taking into account social aspects like user personality and interpersonal trust, which are capable of affecting the item selection process during interactions. To consider such important factors, we propose in this paper a novel approach to group recommendations based on fuzzy influence-aware models for Group Decision Making. The proposed model calculates the influence strength between group members from the available information on their interpersonal trust and personality traits (possibly estimated from social networks). The estimated influence network is then used to complete and evolve the preferences of group members, initially calculated with standard recommendation algorithms, toward a shared set of group recommendations, simulating in this way the effects of influence on opinion change during social interactions. The proposed model has been experimented and compared with related works. • Group recommender systems suggest items to be consumed by groups of users. • Individual preferences of any group member must be taken into account. • Considering social influence between group members leads to better recommendations. • Influence strength can be estimated from information retrieved in social networks. • Fuzzy decision making allows the influence-based aggregation of individual preferences. [ABSTRACT FROM AUTHOR]