Analýza heterogenních informačních sítí

Abstract

The goal of this master thesis is to familiarize with and explore the methods used for heterogeneous information network analysis. We’ve selected and processed a suitable dataset, which can be represented by a heterogeneous network. We’ve created a tool to manipulate the dataset, which will allow a user to select a subset intended for analysis. We’ve explored meta path semantics that can be created from this data. Suitable meta paths were used to create a bipartite graph or a homogeneous network through projection. A part of movie database IMDb was used as input for the implemented ranking algorithms. These algorithms are used for homogeneous or heterogeneous network analysis and provide a distinct view into the network. Algorithm outputs were visualized and interpreted.