Modeling data heterogeneity using big dataspace architecture

Abstract

With the wide use of information expertise in advanced analytics, basically three characteristics of big data have been identified. These are volume, velocity and variety. The first of these two have enjoyed quite a lot of focus, volume of data and velocity of data, less thought has been focused on variety of available data worldwide. Data variety refers to the nature of data in store and under processing, which has three orthogonal natures: structured, semi-structured and unstructured. To handle the variety of data, current universally acceptable solutions are either costlier than customized solutions or less efficient to cater data heterogeneity. Thus, a basic idea is to, first design data processing systems that create abstraction that covers a wide range of data types and support fundamental processing on underlying heterogeneous data. In this paper, we conceptualized data management architecture ‘Big DataSpace’, for big data processing with the capability to combine heterogeneous data from various data sources. Further, we explain how Big DataSpace architecture can help in processing the heterogeneous and distributed data, a fundamental task in data management.