A novel group recommender system based on members' influence and leader impact

Group recommender systems have been designed which, instead of suggesting one or more items to people individually, concurrently recommend them to a group of people who have a common interest, with a view to satisfying each of them. One of the most important issues in these systems is social relationships and the influence of individuals on each other in groups. In this article, a new method has been proposed to compute members' influence on each other based on similarity and trust. Normally in groups, there are some people called Leaders who are trusted more than other members and have a significant impact on the members. Therefore, this study has attempted to compute the leader's impact on the members' preferences. One remarkable aspect of this method is the use of a combination of fuzzy clustering and similarity measure to find users who have similar interests. Furthermore, an implicit trust metric has been formulated to improve the efficiency of the influence process and leader identification. Eventually, the proposed method which has been evaluated utilizing a MovieLens 1001k dataset showed significant results by MAE, RMSE, Precision, and a group-satisfactionmeasure compared to state-of-the-art techniques. Further, the proposed trust metric has shown better efficiency compared to some state-of-the-art methods. (C) 2020 Elsevier B.V. All rights reserved.