ProfKaska Musial-Gabrys

My main areas of research are complex networked systems, and analysis of their dynamics and evolution, as well as predictive, adaptive modelling of networked systems. I also recently started research in a new direction - the application of machine learning approaches to networked, dynamical systems. Perfect example of such systems is social network, a concept that we all know very well as each of us is a part of one global network. This network is created by people and the interactions between them. We constantly create connections both in the real world (at home, school, office) and in the rapidly growing online world (Facebook, YouTube, Twitter, Flickr). In my research Iinvestigate those systems, their characteristics and how they change over time. Examples of very interesting questions worth investigating are e.g. what causes that when we work together we can achieve more than when we work individually (concepts known as collective intelligence and emerging behaviour) or what makes that some of the videos, pictures, stories spreads through social network so quickly (known as viral chains).

a)Intelligent analysis of large complex networks - the networks that are in the area of my interest are extracted from large datasets obtained from telecommunication companies (British Telecom plc - BT), e-mail servers (WUT, Enron), multimedia sharing systems (Flickr), etc. The first research on investigating and analysing social networks was conducted as part of the EU FP6 Coordination Action project on Nature-inspired Smart Information Systems (11/2005 - 01/2008) where I acted as a member of the Nature-inspired Data Technology focus group. I presented the research results at the NiSIS symposia at Majorca (06/2006), Tenerife (11/2006), and Malta (11/2007).

b)Dynamics and predictive modelling of complex networked systems - this is area that currently becomes the main field of my research efforts. The conducted research is concerned with discovering patterns in nodes' behaviours and the interactions between them. The analysis of these patterns and their changes in time allows prediction of the future behaviour of nodes and their relations. One of the ways to model the network dynamics is the application of methods based on the molecular modelling concept and other physically-inspired methods. Another approach that I investigate is the application of machine learning methods to infer and predict the future structure and characteristics of network.

c)Network motifs method in social networks - Network motifs are small subgraphs that reflect local network topology and were shown to be useful for creating profiles that reveal several properties of the network. The outcomes of my research have revealed that motif analysis enables the effective investigation of both network structure and patterns of interactions between nodes within the network. In addition, the analysis of network motifs dynamics can be utilized in detecting and exploration of changes in complex network structures.

d)Multirelational social networks - these are the networks in which more than one type of relationship exists. Different types of relationships can emerge from various communication channels, i.e. based on each communication channel separate relation that can be also called a layer of a network is created. The relationships are extracted from the users activities and if in the system the knowledge about more than one kind of activity is gathered then more than one type of connection can be defined. Different layers can be also built upon various nature of the connections between users, e.g. co-workers, family members, friends. The systems that can be used in such analysis are the multimedia sharing systems such as Flickr or YouTube, which are typical examples of Web 2.0 systems. In my research I have investigated such systems as Flickr, Vimeo, ExtraDom, and recently Badoo.

e)Evaluation of a user position in a social network - during my PhD I conduct research on assessing a position of individuals in networked systems. The position is calculated based on the users activities and their interactions. The method and appropriate algorithms were developed and number of experiments on real-world networks was carried out. The research on evaluating user position was supported by the individual grant that I obtained from the Polish Ministry of Science and Higher Education (06/2008 - 12/2009).

f)Modelling of complex adaptive software systems - the main challenge of the current software systems is to build the systems that will be able to adapt to the changing external environment. The research conducted within the Advanced Software Engineering task within INFER project was focused on developing architectures for complex adaptive systems. I worked on that task when I was employed at BU.