Χρονικά μεταβαλλόμενα μέτρα κεντρικότητας και πρόγνωση δεσμών σε κοινωνικά δίκτυα
Time-aware network centrality measure and link prediction
View/
Keywords
Support Vector Machines (SVM) ; Network analysis ; Social mediaAbstract
In network analysis predicting future links is an important task that provides further understanding and knowledge about the evolution of a social network. As social networks are constantly increasing the link prediction has attracted much attention. In this paper, we split the social network in subsets based on the timestamp of each interaction. For each subset the similarity matrices are calculated and using the Support Vectors Machine algorithm we train the estimator. We present the results of the classification method based on accuracy of its prediction between two consecutive subsets.