in course degree Informatics
Supervisors and topics for admission to study
in academic year 2020/21
Doctoral degree studies at Faculty of Informatics and Information Technologies SUT in Bratislava
doc. RNDr. Anna Bou Ezzeddine, PhD. (anna.bou.ezzeddine[at]stuba.sk)
- Metaheuristic optimization
Optimization is often encountered in real life. Entrepreneurs seek to maximize profits, manufacturers expect the highest efficiency of their production processes, investors try to minimize the risk of investment, engineers try to achieve the best performance of their designs. Resources, time, and money are always limited in the real world, so the optimal utilization of these available resources is extremely important. Most real-world optimization tasks are highly non-linear and multimodal, with various complex constraints. Finding an optimal solution, or even a suboptimal solution, is generally not an easy task.
Intelligent meta-heuristic algorithms provide a convenient and practical solution to complex optimization problems. Within metaheuristics, nature-inspired algorithms are gradually coming to the forefront because they are intelligent, can learn and adapt to the environment, much like biological organisms. These algorithms are interesting to the scientific community mainly because of the increasing complexity of the problems that need to be solved. They allow processing tasks with incomplete, probabilistic information when making decisions.
The topic includes the following research subtopics, which will be examined separately.- The area of our energy research will include optimal dynamic predictions of energy production and consumption that will enable real-time planning and management of microgrid operations.
- In the field of bioinformatics, we will focus on research aimed at the ability to search for DNA mutations that cause some selected types of cancer. DNA changes can also be investigated in connection with the patient’s health information.
- In the field of banking, we will explore the possibilities of using nature-inspired algorithms to improve macroeconomic forecasts in Slovakia and the euro area.
- Antisocial Behaviour in Online Communities
Antisocial behavior in online environment is serious problems. As a typical example, we can take the spread of misinformation through social networks, spread of hate speech or direct attacks on persons in online space. The anonymity in the enviroment in information technologies fuels the antisocial behavior. Antisocial behaviour is researched across the whole spectrum of online communities, in the social network sites, news portals, Community Question Answering - CQA systems, online games, online encyclopedias and discussion fora. An actual research challenge is automatic recognition of antisocial behaviour in online communities. - User Model for identification
Biometric features such as keystroke dynamic (dynamic typing), dynamic work with the mouse, the activities carried out in the computer system, dynamic with mobile are characterized by the user. They can be used for user modelling to identify, recognize the emotional state, the detection of computer experience. The potential of examining are for creating user profiles in the computer system, to identify the user to increase the accuracy of user profiles for different types of users. - Trust and privacy in recommendation on the Web
The recommender systems are an integral part of almost every modern Web application. While interacting with various systems, users generate information (in form of various feedback), which we use in the personalized recommendation. Based on the application domain, users tend to distrust the generated recommendations (e.g., news domain, e-commerce). This is a great research opportunity, which is usually addressed by the explanation of recommendations (i.e., explanation why or how the recommendation was generated). Hand in hand with explanation also the visual aspect of explanations is an interesting topic of the actual research. Despite the recommendation explanation, business is interested in the topic of users’ privacy (i.e., generating recommendations which don’t indicate what used bought already). Novel approaches for explanation and privacy help to increase the trust and a loyalty of users, while these are crucial for the business. We expect a direct active participation of student in commerce projects. We welcome interdisciplinary students (e.g., graduate of sociology). - Personalized recommendation and prediction or user’s short-term behavior
Personalized recommendation is an important part of a modern web applications (e.g., e-shops). Thanks to the nature of nowadays users’ interactions on the Web, most of customers are new (usually only a history from one session is available). Clearly, generate recommendations for these users is challenging. Today, researchers focus on the actual user context and intent identification (based on short-term history of the user). Thanks to this, we are able to address various goals, the user tries to fulfill within his/her actual session. As a result, the web application increases user’s satisfaction. Another perspective and research direction represent methods for prediction of such user behavior. This brings various research opportunities as methods for the intent and context identification or novel approaches for personalized recommendations (interdisciplinary graduates are welcomed, e.g., mathematic field of the study). We expect a direct participation of the student in research and industry projects. - Informatic tools for modeling and controling of smart grids
Control of energy networks equipped with intelligent measuring devices - smart grids - is a challenging and complex issue. Smart meters provide a huge amount of data from regular measurements whose processing has great potential and is inevitable for their management. The complexity of the smart grids management is rising with the increasing activity of all stakeholders. Intelligent network models include, in particular, precise models for energy production and consumption and also price models that are important for all energy market participants - vendors, suppliers and customers. A new modern concept is the requirement of flexibility and energy autonomy of smaller units, so called microgrids. These are flexible nodes of the network, containing multiple sources of energy, multiple users of different types, multiple energy storage. It can be viewed as separate business units that can trade with each other, aiming at minimizing energy costs, assuming moving prices. An important role is played by communication with the customer, its active role in controlling the production and consumption of electricity. Modeling involves renewable energy sources, energy storages, weather influence and other parameters. It requires application of special methods of static and stream data processing as well as the use of blockchains. - Optimization and Reuse in Development of Large Software Systems
Increasing requirements and the complexity of designed systems require sophisticated and creative process using intensive collaboration of multiple experts with various specializations, which are informed about the real condition and problems just in time of analysis and construction. Therefore, the main motivation of this research is to analyse and propose the various methods for:- collaborative and parallel development on the shared models, increasing the productivity and work efficiency,
- visualizing parallel layers with particular components, use cases, versions, alternative and parallel flows to reduce the vague and redundant elements to reach the lean and optimal architecture,
- refactoring and reusing of the knowledge sources in the models and the source code,
- VR or XR multidimensional visualization of source code structure, topics and knowledge sources, quality (patterns and anti-patterns), authors and users, evolution, interconnected with the model to reduce the cognitive load and complexity of the exact and large UML models using the layers decomposing system for exploring and deeper understanding that can lead to more effective applications.
- Data Analytics
Nowadays a huge amount of data is generated every day. In order to get value from these datasets, it is necessary to process them effectively. The information, which is hidden in them can be very useful in supporting decision processes, therefore this topic becomes very popular today. Due to various reasons, the traditional methods are not generally usable, therefore new approaches have to be found.
An actual research area is the optimization in energy domain, but the subject of interest is data analysis in broader context - including the tasks of predictive modeling, clustering, classification or anomaly detection in various domains. Interesting challenges are particularly those tasks, where the processing depends not only on the sequence of events, but also on the broader knowledge context.
The PhD project will be linked to the research project, or the cooperation with industry. - Advanced recording and modeling of user behavior
Recording and modeling of user behavior brings yet more challenges: how to utilize data from specialized sensors (e.g. eye-trackers), how to better utilize contextual information, how to transfer the models between domains (tasks), how to represent the knowledge about the user or if it is desirable to do it at all (e.g., neural networks). Possible domains of applications related to research project at the faculty: consumption and spread of misinformation on the web, technology enhanced learning, map applications, crowdsourcing. - Text mining and information extraction
Increasing amount of available information, emphasized by incessant expansion of the Web, naturally accentuated the information overload problem.
It is necessary to discover and develop new approaches to text data processing for applications allowing more efficient and valuable access to information (in both broader and narrower sense). Machine learning, statistical processing and recently also models of neural networks gain on importance, yet many questions regarding text processing remain unanswered. It is necessary to seek new or improve existing models representing text and information stored within, eventually focusing on domain specifics. It is inevitable to research new and improve existing methods for text mining, natural language processing or information extraction. Nature of the text being processed, its quality or language used may constitute potential pitfalls to deal with.
Seeking for new knowledge within this research topic is rooted in practical applications of both research and industry projects. Current research challenges include sentiment analysis, text summarization, fake news detection, antisocial behaviour detection and information extraction from user generated content. - Models of adaptive web-based systems
In various scenarios of accessing information on the web we face the so called information overload problem. This applies for both open and closed information systems.
Users on the web need - and today even expect - mechanisms for improving and simplifying access for information. Information systems became intelligent in a sense of adapting the content and/or connections (infrastructure) of information to users based on their goals in the information system. The research of methods and techniques for adaptation continues to be very relevant today. The number of information systems (both kinds and instances) potentially benefiting from adaptation increase. Educational web-based systems are especially challenging. Despite more than two decades of research many research questions remain not answered. They particularly cover different models of an educational information system (domain model, educational model, user/student model, navigation model, model of goals, etc.). It is important to research novel models and methods for adaptation in web-based information systems in order to address the needs of web users. - Patterns in Agile Software Development and Beyond
Effective ways of organizing people known from agile software development have been captured in the form of organizational patterns (e.g., Architect Also Implements, Engage Customers, or Developer Controls Process). There is a need to make organizational patterns more accessible to all the people involved in software development, to facilitate their incorporation in organizations, and to explore how they are related to software modularization. Here, peripheral perception as a means of highly effective and effortless communication among developers deserves a particular attention. On a more general level, the possibilities of transposing the idea of patterns, as affirmed in software development, to other areas, including social innovation and arts, should be explored. - Representing and Reusing Software Knowledge
Enormous intellectual efforts are being invested into producing software, but only a fraction of all the software knowledge being created in this process is readily reusable. Further research of software product lines and variability management, domain specific languages, and software patterns could help change this. Speaking more generally, it would be worthwhile exploring how heterogeneous software knowledge contained in various artifacts being produced during software development (including numerous documents and informal sketches) can be efficiently represented and reused. One particularly interesting idea is to interrelate and visualize software knowledge employing multidimensional software modeling, in which domain specific modeling combined with multilevel metamodeling might help. Software Comprehension
Software is not only difficult to create, but it is also difficult to understand. Appropriate ways of preserving the comprehensibility of the intent with which software has been developed and how to achieve this directly in the corresponding code and models should be explored. Part of the solution lies in establishing and maintaining appropriate modularization of code and models, such as use case driven modularization or aspect-oriented modularization. However, no modularization is suitable for all occasions, generating a need for a seamless switching mechanism from one modularization to another, which might be seen as a form of a continuous and automated refactoring. The topic of software comprehension is much wider and it extends all the way to software testing and education in software development.
doc. Mgr. Daniela Chudá, PhD. (daniela.chuda[at]stuba.sk)
doc. Ing. Michal Kompan, PhD. (michal.kompan[at]stuba.sk)
prof. RNDr. Mária Lucká, PhD. (maria.lucka[at]stuba.sk)
doc. Ing. Ivan Polášek, PhD. (ivan.polasek[at]stuba.sk)
doc. Ing. Viera Rozinajová, PhD. (viera.rozinajova[at]stuba.sk)
doc. Ing. Jakub Šimko, PhD. (jakub.simko[at]stuba.sk)
doc. Ing. Marián Šimko, PhD. (marian.simko[at]stuba.sk)
doc. Ing. Valentino Vranić, PhD. (valentino.vranic[at]stuba.sk)