As one of the most impactful emerging technologies, big data analytics and its related applications are powering the development of information technologies and are significantly shaping thinking and behavior in today’s interconnected world. Exploring the technological evolution of big data research is an effective way to enhance technology management and create value for research and development strategies for both government and industry. This paper uses a learning-enhanced bibliometric study to discover interactions in big data research by detecting and visualizing its evolutionary pathways. Concentrating on a set of 5840 articles derived from Web of Science covering the period between 2000 and 2015, text mining and bibliometric techniques are combined to profile the hotspots in big data research and its core constituents. A learning process is used to enhance the ability to identify the interactive relationships between topics in sequential time slices, revealing technological evolution and death. The outputs include a landscape of interactions within big data research from 2000 to 2015 with a detailed map of the evolutionary pathways of specific technologies. Empirical insights for related studies in science policy, innovation management, and entrepreneurship are also provided.

Introduction

It has been several years since the big data boom led a revolution in both re-shaping thinking and behavior in all sectors of modern society (Mayer-Schönberger and Cukier, 2013). The broad range of big data applications in business intelligence has been highlighted by both academic and business communities (Chen et al., 2012). Defined as “the means of managing, analyzing, visualizing, and extracting useful information from large, diverse, distributed, and heterogeneous data sets1 ”, big data analytics has become essential for the success of commerce (McAfee et al., 2012). Such increasing significance of big data, of course, attracts the attention of science, technology, and innovation policy (STIP) communities, and bridging big data with real-world STIP issues has been raised as an urgent task – e.g., investigating the potential (including both positive and negative impacts) of big data and providing feasible reactions for specific industry sectors (Kwon et al., 2015; Marx, 2013; Nobre and Tavares, 2017). Unfortunately, despite the ambition of big data analytics on creating “big impact” (Chen et al., 2012; Wamba et al., 2015), the success of big data analytics in non-IT companies is still limited (Court, 2015). Now is an opportune time to trace the evolution of big data research to discover the interactions between the techniques used in big data analytics and identify the crucial connections that have the potential to create and extend the sphere of “big impact”.