فهرست نشریات علمی دسترسی آزاد ایران

 In this paper we present a novel distributed link prediction method for social networks that is scalable and usingstructural properties of the network for its predictions without need to any profile. This is an agent-orientedmodeling which doing good community detection and using the gravitational search algorithm to select candidatelinks between communities. The experimental results of prediction on various data sets show that the proposedmethod is scalable with on average 69% precision and 68% accuracy. Also, better scalability with improvement inspeed, precision and accuracy has been reported by selecting an appropriate or optimize CPU allocation.