Big Data: Methodological Challenges and Approaches for Sociological Analysis

Abstract

The emergence of big data is both promising and challenging for social research. This paper suggests that realising this promise has been restricted by the methods applied in social science research, which undermine our potential to apprehend the qualities that make big data so appealing, not least in relation to the sociology of networks and flows. With specific reference to the micro-blogging website Twitter, the paper outlines a set of methodological principles for approaching these data that stand in contrast to previous research; and introduces a new tool for harvesting and analysing Twitter built on these principles. We work our argument through an analysis of Twitter data linked to political protest over UK University fees. Our approach transcends earlier methodological limitations to offer original insights into the flow of information and the actors and networks that emerge in this flow.

The adoption of Big Data Services by Manufacturing firms: An empirical investigation in India.

“…Shin (2014) defined data quality as the fitness for use of information, determined by whether or not the data meets the requirements of its authors, users and administrators. Big data is less a matter of data volume than the quality of data to improve quality and efficiency in the delivery of services (Kwon et al, 2014;Tinati et al, 2014). The four core point of the big data quality are (Gandomi and Haider, 2015) 1.…”

Challenging social media analytics

“…Flow 140 (described in detail in [44,45]) is a new network analytics platform built on the well-established techniques and metrics developed in social network analysis (SNA) studies, adjusted to explore the emergence of information flows and network roles over time. Following the sociology of networks, mobilities and flows Flow 140 is distinguished from conventional SNA in three key ways.…”

“…Using Flow 140 for a case study of the use of Twitter in political protest [45] revealed which users were key to the generation and flow of information and the different types of roles that were involved. These stretch beyond quantitative measures of re-tweets to include 'amplifiers' and 'aggregators' who -whilst not necessarily highly retweeted themselves play an important role in the diffusion of information and in building connections between discrete networks.…”

Micro-text classification between small and big data

“…Analyzing these micro-texts has a great scientific potential in psychology [3] and social sciences [4], but is methodologically challenging [5] and typically heavily depends on machine text processing to cope with the large amounts of published texts. A fundamental step in social data mining is text classification, which may involve both identifying texts that match predefined classes (top-down classification) as well as finding an unknown classification scheme that fits the data (bottom-up classification).…”