Lecture is about network dynamics, briefly, analysis of the changes that occur
in the network in time. These changes include appearance and disappearance of nodes,
links and also some aggregate metrics of the network. In addition to the analysis
of the known changes, network dynamics deals with predictions about future changes
in the network, like formation and dropping of new edges.
First study mentioned in the class was "Empirical Analysis of an Evolving Social Network".
Paper analyzes a university e-mail network. Ties are weighted with the number of messages
within the last 60 days between the pairs. One major result is that, individuals sharing at
least one class are 3-times more likely to start emailing each other if they have an email contact
in common. Probability significantly increases as number of ties or shared classes increase.
Next part was about the Team Assembly Mechanisms and Evolution of Affiliation Network
Related to Social Network. Examples studies in this field are Broadway Musical Industry and
Collaboration Networks in different sciences. Interesting findings can be summarized as follows;
If your friends join, so will you.
Cliquish communities grow more slowly.
Second part of the lecture was about the results of the affects described in the first.
As the nodes and links change over time, how does the degree distribution,
clustering coefficient and average shortest path change. Graph Densification is analyzed
for physics citations, patent citations, autonomous systems and affiliation networks.
A mathematical model is given. The main result is that number of edges are growing faster
than nodes, so average degree is increasing. Similar studies are carried out for the
diameter change.
In the final part, possible explanation of densification, community structure and difficulty of cross linking, and forest fire model is discussed.
No comments:
Post a Comment