Efficient -means++ Approximation with MapReduce

Abstract— Efficient -means++ Approximation with MapReduce. The important person is vital for information spreading efficiency in social networks, which has always been the hottest research field in recent years. Nowadays, with the increased < Final Year Projects > information spreading speed in our daily life such as cellphone or microblog, the government and enterprise have realised that it is necessary and important to manage and control the information spreading in social networks of people by recognising the important person or VIP customer. Meanwhile, with the rapid increase of information in society activities recently, the distributed information system engineering such as MapReduce-based computation system has become more and more popular in large-scale social network data analysis in emergency case management or VIP customer recognition. In this article, a novel effective and efficient node betweenness approximation pivot selection method for largescale graphs is proposed based on MapReduce-based system for social network analysis (SNA). Furthermore, it has been proved useful in four discrete datasets and one continuous dataset (six month cellphone call graph) in VIP customer discovery for large graphs from real telecom data in China to help deal with customer relationship management more efficiently.