Understandable Big Data: A survey

Abstract

This survey presents the concept of Big Data. Firstly, a definition and the features of Big Data are given. Secondly, the different steps for Big Data data processing and the main problems encountered in big data management are described. Next, a general overview of an architecture for handling it is depicted. Then, the problem of merging Big Data architecture in an already existing information system is discussed. Finally this survey tackles semantics (reasoning, coreference resolution, entity linking, information extraction, consolidation, paraphrase resolution, ontology alignment) in the Big Data context.

Introduction

Today, people and systems overload the web with an exponential generation of huge amount of data. The amount of data on the web is measured in exabytes (10¹⁸) and zettabytes (10²¹). By 2025, the forecast is that the Internet will exceed the brain capacity of everyone living in the whole world  [1]. This fast growth of data is due to advances in digital sensors, communications, computation, and storage that have created huge collections of data.¹  The term Big Data had been coined, by Roger Magoulas (according to  [2]), to describe this phenomenon.

Seven recent papers (including  [3] and  [4]) have aimed to extract Big Data trends, challenges and opportunities.  [5] provide a survey on scalable database management: updating of heavy application, analytics and decision support. Likewise,  [6] study analytics in Big Data with a focus on data warehouse. These two papers have different goals comparatively to  [7]. In a more rigorous way, M. Pospiech and C. Felden  [7] have selected relevant and recent papers which tackle different aspects of Big Data and have clustered them in four domains: Technical data provisioning (acquisition, storage, processing), Technical data utilization (computation and time complexity), Functional data provisioning (information life cycle management, lean information management, value oriented information management, etc.) and Functional data utilization (realms where big data is used). At the end of their clustering,  [7] note that a lot of papers (87%) are technical and that there is not any paper on functional data provisioning. More closed (compared to the three previous works) to our target, semantics in the age of Big Data,  [8] focus on knowledge discovery and management in Big Data era (flooding of data on the web). As our paper they zoom on gathering relational facts, information extraction, emergence of structure, etc. But a deep circonscription of the concept of Big Data is not in the scope of their article like some other key themes of this paper like reasoning on large and uncertain OWL triples, coreference resolution, ontology alignment. The last paper has been authored by  [9]. They present Big Data integration in a easy-understandable-way. Schema alignment, record linkage and data fusion are presented w.r.t to Big Data characteristics (volume, velocity and variety). Knowing the high value carried by data in general and thus by Big Data, it is not surprising therefore that Chief Information Officers (CIOs) are interested in it analytics as technological. If initially web pages and traditional databases were the raw materials respectively for search engine companies and other businesses, now it has been mixed with large sets of miscellaneous, heterogeneous and unstructured data. It implies that tools and techniques have to be designed to disambiguate it before putting it together to master and manage data of organizations. Our work is similar to  [9] in the approach. We discuss challenges and opportunities of semantics in the age of Big Data and present the supply chain to handle it. Therefore, this article defines Big Data (Section  2), briefly discusses its management (Section  3) and finally tackles Big Data and semantics challenges and opportunities (Section  4).