Social Network Analysis


The last decades have been marked by the rise of online social networks as the primary medium for people to express themselves and connect with each other. In this project we study different problems that emerge from the analysis of the data produced on these social networks.

An important line of work centers around opinion formation processes and polarization on social networks. The question of how opinions are formed in a social network has attracted considerable attention in different scientific disciplines. There are different mathematical models that capture the mechanisms of opinion formation process. Building on these models, we are interested in understanding phenomena such as polarization of opinions, and devising algorithms that gear the network towards a desired state. In [1] we consider the problem of selecting a set of nodes to be positive so as to maximize the average opinion in the network. In [2] we provide a measure for polarization in the network, and propose algorithms for moderating polarization by convincing nodes to take a neutral stand.

Another important mechanism in online social networks is link recommendations. Link recommendations are produced by the network owners as a means to augment the ego-network of individual users and increase their engagement. However, they can also be applied as a way to affect the properties of the network as a whole. For example in [3] we considered the problem of recommending links so as to increase the centrality of the nodes.

Other important problems in Social Network Analysis include: Understanding the strength of ties in social networks [4]; Creating teams of social network users that have the skills to complete a task and are also well connected; Dealing with networks that have both friend and enemy relationships.


[1] Aristides Gionis, Evimaria Terzi, Panayiotis Tsaparas. Opinion Maximization in Social Networks. SDM 2013

[2] Antonis Matakos, Evimaria Terzi, Panayiotis Tsaparas. Measuring and moderating opinion polarization in social networks. Data Min. Knowl. Discov. 31(5): 1480-1505 (2017)

[3] Nikos Parotsidis, Evaggelia Pitoura, Panayiotis Tsaparas. Centrality-Aware Link Recommendations. WSDM 2016

[4] Stavros Sintos, Panayiotis Tsaparas. Using strong triadic closure to characterize ties in social networks. KDD 2014