In recent years, the rise of misinformation and polarization within social networks have made headlines worldwide. The ability to detect and analyze social network communities can provide valuable insights into how information spreads and how different communities interact. This can help develop strategies to mitigate the impact of misinformation and polarization.
Professor Hovhannes A. Harutyunyan and PhD student Sahar Bakhtar in the Department of Computer Science and Software Engineering at Concordia’s Gina Cody School of Engineering and Computer Science have published a conference paper detailing the development of a cutting-edge local community detection algorithm called AlgSP. This innovative algorithm is designed to efficiently detect communities in signed networks.
Signed networks are a type of social network where relationships members are characterized by either positive (friendly) or negative (unfriendly) connections. These connections are referred to as links.
To evaluate the performance of AlgSP, the researchers compared it to five other algorithms on six different datasets, including real-world networks such as the U.S. Supreme Courts and politicians in the U.S. Congress. AlgSP achieved the best possible results out of five of the six datasets. The algorithm also proved to be among the fastest in community detection.
The research has the potential to greatly enhance our understanding of complex social networks and improve various applications, such as recommendation systems and social network analysis.
Read the 2022 Ninth International Conference on Social Networks Analysis, Management and Security (SNAMS) paper “A Fast Local Community Detection Algorithm in Signed Social Networks” added to IEEE Xplore on March 15, 2023.