Complex Temporal Networks: Metrics and embeddings for problems in real world complex networks

TY - BOOK

T1 - Complex Temporal Networks

T2 - Metrics and embeddings for problems in real world complex networks

AU - Ryther, Christopher Alan

PY - 2018

Y1 - 2018

N2 - Temporal Graphs are collections of nodes and edges which change over time. These types of graphs can model many types of real-world scenarios such as email communications, online social networks, author citation networks and so on. With the growth of data globally, graph data has become more complex, e.g. introducing node attributes and labels, therefore new tools and methods are needed to efficiently analyze and solve complex graph problems. This thesis studies several challenges in static and temporal graphs and aims to explore how to best utilize complex network data.The thesis consists of four parts. The first part of the thesis describes how to model both static and temporal graphs, as well as attributes and labels, in a streaming fashion, and important graph metrics and properties of nodes. The conclusions of this thesis are that the use of node attributes and labels, combined with additional temporal information, can improve accuracy when solving classification tasks in temporal graphs. Furthermore, it shows that graphs can be summarized effectively with the use of mixtures of distributions when enough data is present. While examples of applications of the proposed methods have been given, there are still many interesting challenges ahead

AB - Temporal Graphs are collections of nodes and edges which change over time. These types of graphs can model many types of real-world scenarios such as email communications, online social networks, author citation networks and so on. With the growth of data globally, graph data has become more complex, e.g. introducing node attributes and labels, therefore new tools and methods are needed to efficiently analyze and solve complex graph problems. This thesis studies several challenges in static and temporal graphs and aims to explore how to best utilize complex network data.The thesis consists of four parts. The first part of the thesis describes how to model both static and temporal graphs, as well as attributes and labels, in a streaming fashion, and important graph metrics and properties of nodes. The conclusions of this thesis are that the use of node attributes and labels, combined with additional temporal information, can improve accuracy when solving classification tasks in temporal graphs. Furthermore, it shows that graphs can be summarized effectively with the use of mixtures of distributions when enough data is present. While examples of applications of the proposed methods have been given, there are still many interesting challenges ahead

UR - https://rex.kb.dk/permalink/f/h35n6k/KGL01012053087

M3 - Ph.D. thesis

BT - Complex Temporal Networks

PB - Department of Computer Science, Faculty of Science, University of Copenhagen

ER -