Assessing Strategies for Sampling Dynamic Social Networks

Social Networks represents an invaluable source of information to detect, understand and predict social trends and complex dynamics. Unfortunately, the presence of several constraints in data collections as costs, dimensions, time and so forth, requires the implementation of a sampling strategy able to maximize the information value for the analysis. The paper defines a number of parameters and thresholds defining a new strategy for data sampling in social network and compares samples obtained with different strategies. The test case has been conducted on the social network of the mayors of four Italian metropolitan areas. Results of the assessment reveal that the parameter designed for configuring a strategy impacts on the dimension, the extraction time and the quality of the generated network as expected. The best tradeoff between quality and execution time has been identified and discussed.