Algorithmic Approaches to Overlapping Community Detection – Clustering Graphs into Subgraphs

April 11th, 2022

Thesis Type Bachelor
Student Jan Mortell
Status Running
Supervisor(s) Ralf Klamma

(Social) Network Analysis is investigating (social) structures of real-world networks. Networks are composed of nodes and links. Communities are sub-networks whose nodes have more links to nodes within the sub-network than to nodes outside the sub-network. Overlapping Community Detection is the problem of identifying nodes in networks that belong to more than one sub-network. The overlapping community detection problem has an enormous importance for different fields of science like biology, neurology, sociology, media science, politics, economics, and computer science. A vast number of papers has been written about various aspects of overlapping community detection like measures of quality of overlapping communities, modeling, visualization. A huge number of algorithms have been proposed. Recently, methods of machine learning and quantum computing have been applied on the problem. This bachelor thesis should research recent algorithmic approaches to community detection algorithms. Results should be integrated in the existing award winning WebOCD framework, a collection of Java-based microservices deployed in a peer-to-peer network. With the help of the framework, an interactive online book about overlapping community detection algorithms has been started for use in teaching. Open requirements of the WebOCD framework are available in our Requirements Bazaar. Interests in desk research, formal modeling, and Web programming are prerequisites for this bachelor thesis.

Related Projects: