Analýza struktur v komplexních sítích zaměřená na posouzení kvality vybraných metod

Abstract

Complex networks are often used in data analysis. By analyzing the structure of the network, we can obtain a lot of important information and many methods and algorithms are used for different purposes. The aim of this work is to perform experiments with selected algorithms and compare their results. One of the classic tasks is finding communities on the network. In the first part of this work, we used two methods of detecting overlapping communities. Our goal was to compare and evaluate the quality of the found communities on the basis of three measures. Networks can also be used to analyze vector data. First, however, it is necessary to construct a network from vector data using the selected method, but it is not always clear which would be the most appropriate to use. The second part of the work describes the performed experiment, in which we used four methods of construction of networks from vector data and then described the properties and differences of the obtained networks. We will also use the so-called similarity purpose function described in more detail in this work to evaluate the quality of constructed networks.