Community Detection on Social Networks (Bachelor Thesis, Finished)


Paul Höft


Community detection has become an essential tool for the analysis of large data in the age of social networks. Suggestion algorithms, rumor analysis and the detection of echo chambers are not possible without analyzing large data sets. In order to analyze the required amount of data fast and precise, high-performance algorithms and tools are needed that scale well and still deliver sufficiently good results. In this thesis, we compare different algorithms on Zachary’s Karate Club, a social interaction network, Twitter data and Reddit Data for their scalability and the quality of their results. For this purpose, we have collected datasets from a variety of different sources. We analyzed them with different algorithms using a toolstack with proven software and compared the quality of these results with metrics. Our Results show that Louvain Modularity combined with Neo4j is able to tackle large datasets and shows the limitation and bottlenecks of other algorithms.

Start / End Dates

2019/02/20 - 2019/05/20


Research Topics