Analýza heterogenních sítí

Abstract

The aim of this master thesis is to introduce the heterogenous network analysis and data mining tasks, which are used in this research area. This includes: selecting suitable dataset, which could be represented as a heterogenous network, ensuring manipulation with this data, introducing possibilities of selection and filtering of a subset from the dataset, which would be relevant for the analysis purpose, and enabling by that working with smaller dataset. Exploring semantics of metapaths, which can be constructed from this data. Using the metapaths as a way of valuation of edges in homogenous network constructed via projection, or heterogenous network with two types of objects, represented by bipartite graph. Selection and implementation of clustering algorithms, which are based on different principles for these network concepts and enabling by that different views on these networks. Finally using these algorithms for analysis of different metapaths, interpreting and vizualizing their results.