Start Date
11-8-2016
Description
Detecting communities of interest in social media platforms provides insight into the platforms and the individuals that use them. The bulk of research in community detection is directed at network analysis of individuals and their interaction with other members within the network. However, connections outside the network can also be useful for community detection, as in the following of elite Twitter users by regular users. This research develops a mechanism for clustering elite Twitter users on the basis of connections and interactions within their followers. Since clustering is sensitive to initial configurations, the approach is modified using genetic algorithms to traverse multiple regions of the solution space. Application of this approach to a set of 25,000 Twitter users demonstrates that it forms coherent communities within a few iterations, outperforming other clustering approaches for community detection.
Recommended Citation
Sharif Vaghefi, Mahyar and Nazareth, Derek, "Detecting Communities of Interests in Social Media Platforms using Genetic Algorithms" (2016). AMCIS 2016 Proceedings. 3.
https://aisel.aisnet.org/amcis2016/Decision/Presentations/3
Detecting Communities of Interests in Social Media Platforms using Genetic Algorithms
Detecting communities of interest in social media platforms provides insight into the platforms and the individuals that use them. The bulk of research in community detection is directed at network analysis of individuals and their interaction with other members within the network. However, connections outside the network can also be useful for community detection, as in the following of elite Twitter users by regular users. This research develops a mechanism for clustering elite Twitter users on the basis of connections and interactions within their followers. Since clustering is sensitive to initial configurations, the approach is modified using genetic algorithms to traverse multiple regions of the solution space. Application of this approach to a set of 25,000 Twitter users demonstrates that it forms coherent communities within a few iterations, outperforming other clustering approaches for community detection.