Relationship Analysis of Network Virtual Identity Based on Spark

摘要

With the development of Internet, there are massive network accounts on Internet when people tend to contact with each other by network applications. We define these network accounts as Network Virtual Identities (NVIs). According to previous methods, we will propose a method based on Spark to analyze the relationships between NVIs. In order to analyze relationships and handle large-scale data, our approach will implement the parallelized overlapping community detection algorithm COPRA, based on GraphX, a graph-parallel computation framework on spark. Then we develop a strategy to divide the relationships to different strength levels.