ABSTRACT
D ata management and analysis has stimulated paradigm shifts in decision-making in various application domains. Especially the emer-
gence of big data along with complex and social networks has stretched the imposed requirements to the limit, with numerous and crucial potential benefits. In this chapter, based on a novel approach for big data analytics (BDA), we focus on data processing and visualization and their relations with complex network analysis. Thus, we adopt a holistic perspective with respect to complex/social networks that generate massive data and relevant analytics techniques, which jointly impact societal operations, e.g., marketing, advertising, resource allocation, etc., closing a loop between data generation and exploitation within complex networks themselves. In the latest literature, a strong relation between hyperbolic geometry and complex networks is shown, as the latter eventually exhibit a hidden hyperbolic structure. Inspired by this fact, the methodology adopted in this chapter leverages on key properties of the hyperbolic metric space for complex and social networks, exploited in a general framework that includes processes for data correlation/clustering, missing data (e.g., links) inference, social network analysis metrics efficient computations, optimization, resource (advertisements, files, etc.) allocation and visualization analytics. More specifically, the proposed framework consists of the above hyperbolic geometry based processes/components, arranged in a chain form. Some of those components can also be applied independently, and potentially combined with other traditional statistical learning techniques. We emphasize the efficiency of each process in the complex networks domain, while also pinpointing open and interesting research directions.
