Vzorkování síťových dat

Abstract

This master’s thesis describes sampling network data and methods, which are able to sample real networks. The main goals of this thesis are experiments on various network data sets, which are sampled by the described methods and evaluation of experimental results. Graph Sampling is technique to pick a subset of vertices or edges from original graph. In some cases, the whole graph is known and the goal of sampling is to obtain a smaller sample by the methods based on vertex or edge selection. In other cases, the graph is unknown and the sample is obtained by the methods based on graph exploration. There are many algorithms to compute various measures of graphs, which are computationally expensive. This is the reason to create sample from large original graph and run a computationally expensive algorithm on this sample. The sample must preserve the properties of the original graph. The observed properties of the graph and process of measuring the success of the sampling methods based on the comparison of the properties are also described in this thesis.