Improving Similarity Measures for Sampling Social Networks by Eliminating Isolated Nodes

Examining social networks has extensive uses in gathering information and models in various scientific researches. The main obstacle in the study of such networks is the large number of users and their complex interconnections, which makes analysing these networks almost impossible in their real proportions. Hence, the researchers have tried sampling methods of social networks. There are various sampling methods for social networks. In some of these sampling methods, isolated nodes are created which affects the evaluation results. Although each isolated node represents a population of main network nodes, but since each node s selection is important and can change the parameters of the evaluation, it is necessary to examine the selection of subnetwork nodes so that the sampling of these networks can be done in the most efficient manner, and the parameters for assessing the similarity of the networks show the best results. In this research, in a sampling method, the effect of the presence and deletion of the isolated nodes in the sampling is examined and are compared in a number of similarity assessment criteria.