Abstract
dc:descriptionNowadays, networks are used to represent data in almost any field such as biology, transportation, marketing and social groups. Networks comprise a convenient way to study data and the associated relationships among them, via the field of network analysis. In networks, certain groups of entities with particular importance arise, which form the so- called communities. Communities are of particular importance, since they are related to functional units of the network. Consequently, community detection, constitutes a significant research field. To this end, the present thesis confronts three community detection related problems. More precisely, the first part of the thesis deals with local community detection where the process for identifying groups of entities starts from important entities called hints. This additional information about the importance of some entities can be exploited for detection of better and more relevant communities and it is derived from an expert on the domain, or may arise from the network’s side information. Hence, Hint Enhancement Framework is proposed, which applies a two-step procedure to discover the community of the hints. The last part of the first chapter of the thesis deals with the problem of community detection in dynamic networks. In many cases, looking at the evolution of a particular community or of the community that an important node belongs constitutes an interesting network problem. However, it is difficult to keep track of the evolving communities because of the drifting problem, where a community might shift to a completely different one. To confront this drawback, a framework for detecting the community that contains a set of nodes of particular importance, termed anchors, as well as its evolution over time is suggested. This framework, called Dynamic Community Detection with Anchors (DCDwA), circumvents the identity problem by letting the anchors define partially or completely the core of the corresponding community. The second part of the thesis focuses on discovering causal relations between entities and communities, which adds value to the community detection problem in networks. More specifically, a multi- step framework which identifies the causes of an entity participation in a community is introduced. This framework helps in the identification of important edges in a network, which if removed (or added) change some node’s belongingness to a community. Finally, the last part of the thesis concerns the organization and taxonomy of local community detection related work. More precisely, the last part aims to fill a gap and help researchers get a clear overview of the local community detection problem. To this end, along with the taxonomy of the local community detection approaches proposed in the literature mainly in the last decade (2012-2022), the best documented tools and the most commonly used datasets are also gathered to help researchers identify the tools they can use to prove their methods.
Degree
thesis:*- Grantor dc:publisher
- Aristotle University Of Thessaloniki (AUTH)
- Year dc:date
- 2022
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Baltsou, Georgia
- Μπάλτσου, Γεωργία
Subjects
dc:subject × 12- Ανάλυση δικτύων
- Αλγόριθμοι
- Εύρεση κοινοτήτων
- Network analysis
- Algorithms
- Community detection
- Επιστήμη Ηλεκτρολόγου Μηχανικού, Ηλεκτρονικού Μηχανικού, Μηχανικού Η/Υ
- Επιστήμες Μηχανικού και Τεχνολογία
- Τεχνολογία μέσων
- Electrical Engineering, Electronic Engineering, Information Engineering
- Engineering and Technology
- Media Technology
Rights
- Language dc:language
- gre
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/53515