Towards Network Reduction on Big Data

Abstract

The increasing ease of data collection experience and the increasing availability of large data storage space lead to the existence of very large datasets that are commonly referred as "Big Data". Such data not only take over large amount of database storage, but also increase the difficulties for data analysis due to data diversity, which, also makes the datasets seemingly isolated with each other. In this paper, we present a solution to the problem that is to build up connections among the diverse datasets, based upon their similarities. Particularly, a concept of similarity graph along with a similarity graph generation algorithm were introduced. We then proposed a similarity graph reduction algorithm that reduces vertices of the graph for the purpose of graph simplification.