@misc{Kulisiewicz_Marcin_Event-driven_2019, author={Kulisiewicz, Marcin}, contributor={Kazienko, Przemysław. Promotor and Michalski, Radosław. Promotor}, year={2019}, rights={Wszystkie prawa zastrzeżone (Copyright)}, publisher={Politechnika Wrocławska}, language={eng}, abstract={The Universe undergoes a constant change. This affects every subject of our world, and every method used to describe real-world systems should take its dynamics into account. One of the most powerful concepts representing a complex system is a network. In this work, I study temporal social networks that extend static networks by another degree of freedom - time. With this additional dimension, temporal networks are able to model systems' changes capturing individual interactions between network nodes. In particular, I propose a set of new entropy-based measures that are capable of quantifying temporal networks dynamics, alongside algorithms for their iterative and parallel computation. Next, by applying those to real-world cases, I demonstrate that human beings are much more selective in their communication over time. In addition, I show that communication within social communities that exist in examined real-world social networks have different entropy, which can be potentially utilized for group recognition. Furthermore, to provide a meaningful method of comparing temporal socialnetworks of different sizes, I define the normalization method for entropy-based measures. In literature, there is a lack of temporal network models that are able to model humans' cognitive processes. Making it another contribution to the field of temporal network modeling, I fill this gap with CogSNet - a new event-based and cognition-driven temporal network model. It provides reinforcement by discrete events and continuous forgetting mechanism that is capable of modeling human perception and cognition processes of mutual interactions. The CogSNet model significantly outperforms all other reference models commonly used in the literature that was proved on the real-world data set. The model is capable to take into account various interactions as well as the heterogeneous nature of human behavior. Both entropy-based measures and the CogSNet model provide different but complementary information about the dynamics of social networks. This has been shown by means of experimental studies. The following thesis contribution is meant to extend our knowledge about dynamic systems we are part of. And by doing so, to provide to the field new tools to deal with the complexity of reality}, title={Event-driven Temporal Social Networks}, type={rozprawa doktorska}, keywords={network science, social networks, temporal networks, entropy, human cognition}, }