ags:temporalgraphs

People: Lutz Oettershagen

In our recent publication Temporal Walk Centrality: Ranking Nodes in Evolving Networks, we introduced a new centrality measure for temporal networks. Identifying and ranking nodes of web-based social networks and communication networks according to their importance in the dissemination of information are critical and challenging tasks, especially if the considered networks are non-static and of temporal nature.

A temporal network is here defined as a finite set of vertices and a set of temporal edges. In contrast to a static edge, a temporal edge has an additional timestamp that determines when the edge is available in the network. Using temporal networks, we can naturally model communication or contact networks. Hence, temporal networks increasingly gain attention in social network analysis.

Why do we need to consider temporality? To answer this question consider the following simple email network:

The temporal email network shows that Alice sends an email to Bob on Tuesday. Bob sends an email to Carol on Monday. Hence, we can exclude any information flow from Alice to Carol via Bob due to the temporal properties of the network. Below the temporal network, we see the underlying static graph that ignores the temporal information. If we ignore our knowledge of the temporal structure we may come to a wrong conclusion and assume that information could flow from Alice to Carol via Bob. This simple example demonstrates that we have to consider temporal information available in dynamic communication and contact networks to avoid the wrong misinterpretation of causality. Hence, algorithms and analysis need to consider such and further temporal properties, e.g., the frequency distribution of temporal edges.

Our research focuses on algorithms for temporal graphs that are specifically designed to respect these peculiarities.

ags/temporalgraphs.txt · Last modified: 2022/05/04 14:38 by lutz.oettershagen