Abstract—Facebook, Twitter and LinkedIn are the most popular online social networking sites on the Internet. These sites are a powerful mode of sharing, organizing and finding content and contacts. Usage of these sites is increasing so as to provide an opportunity to study the characteristics of online social networking sites at large scale. In this paper work, an attempt has been made to estimate the server load of social networking sites in order to maintain the servers efficiently. In this order, we have gathered the data for three popular social networking sites: Facebook, Twitter and LinkedIn from Internet Libraries. Datasets contain data of 600 cities across the world in terms of Number of users and response time respectively. Further, we have applied Dimension Reduction Algorithm to reduce the datasets for the purpose to attain the meaningful data. Thereafter, we have applied two clustering techniques (K-Means and Agglomerative hierarchical clustering) on these datasets to estimate the load of social networking sites. Results confirm that the clusters which arise from both the techniques contain various number of objects which specify that all the objects (i.e. cities) comes under that particular cluster cover same load to some extent that validate the hypothetical claims and exhibit the effectiveness of our algorithms.

 

Index Terms—online social networks, reduction algorithm, XLStat, K-mean clustering, agglomerative hierarchical clustering

Cite: Deepti Bhagwani, Setu Kumar Chaturvedi, and Kapil Keswani, "Load Estimation of Social Networking Web Sites Using Clustering Technique," International Journal of Electronics and Electrical Engineering, Vol. 4, No. 6, pp. 540-546, December 2016. doi: 10.18178/ijeee.4.6.540-546