Map-Reduce-Based Centrality Detection in Social Networks: An Algorithmic Approach

Abstract

Social network analysis is found to be one of the emerging research directions in the field of data science. This paper mainly concerns with the identification of influential entities with the help of several centrality measures like degree, closeness, and eigenvector centrality. The computational efficiency of analyzing social networks is limited by the size and complexity of the network domain. As the size of the network grows at an exponential rate, it is quite challenging to process the massive network with the help of conventional computing resources. In this manuscript, scalability and complexity issues have been addressed to identify the influential nodes in the network by implementing the algorithm in a distributed manner. The distributed approach has been considered in computing different centrality measures like degree, closeness, and eigenvector. In this paper, the centrality measures have been computed by considering both the local and global structural information. Real-world social networks are observed to follow the power law in both centrality drift and degree distribution. In this work, nodes are ranked based on their importance for different centrality measures. The effectiveness of these algorithms is critically examined through experimentation by using six real-time network datasets.