Community Detection and Analysis of Twitter Social Data

Abstract

"Human is a social animal" this line itself explains the importance of society in one's life. Society brings stability, a medium to express thoughts. Society leads to social interaction which eventually brings thoughtful minds. Humans have the intrinsic nature of analyzing and opinionating things and persons. This keen nature of human has emerged a new field of analysis that is social data analysis. Internet has merged the world today and as a result human social circles have expanded. There are various peculiar social networking sites available on internet, some of them are Facebook, Twitter, LinkedIn and many more. Each maintains accounts of billions of active users and huge amount of data is being produced as a result of interactions over such sites. Hence analyzing this data is a tedious task. But analysis of such online social communities and predicting their behaviour is of great importance for businesses and academics. For our research purpose we have used Twitter as a key medium for social data. This thesis aims to develop a research based application using twitter and R-tool for social data analysis. Further comparison of Community detection algorithm(s) on CPU and GPU technology are performed. We have used Nvidia's CUDA toolkit which provides a possibility of increasing the computational efficiency of Community detection algorithms and metrics.