A Comparative Study of Clustering Algorithms

Abstract

Clustering is a major problem when dealing with organizing and dividing data. There are multiple algorithms proposed to handle this issue in many scientific areas such as classifications, community detection and collaborative filtering. The need for clustering arises in Social Networks where huge data generated daily and different relations are established between users. The ability to find groups of interest in a network can help in many aspects to provide different services such as targeted advertisements. The authors surveyed different clustering algorithms from three different clustering groups: Hierarchical, Partitional, and Density-based algorithms. They then discuss and compare these algorithms from social web point view and show their strength and weaknesses in handling social web data. They also use a case study to support our finding by applying two clustering algorithms on articles collected from Delicious.com and discussing the different groups generated by each algorithm.