Social Network Analysis

Interdisciplinary Approaches and Case Studies

Xiaoming Fu, Jar-Der Luo, Margarete Boos, Xiaoming Fu, Jar-Der Luo, Margarete Boos

• Publisher: CRC Press
• Year: 2017
• Print ISBN: 9780367573782
• eBook ISBN: 9781315352572
• Topic: Computer Science
• Subtopic: Computer Networking
• Index: Computer Science

The book addresses the issue of interdisciplinary understanding of collaboration on the topic of social network studies. Researchers and practitioners from various disciplines including sociology, computer science, socio-psychology, public health, complex systems, and management science have worked largely independently, each with quite different principles, terminologies, theories. and methodologies. The book aims to fill the gap among these disciplines with a number of the latest interdisciplinary collaboration studies.

I METHODOLOGIES FOR INTERDISCIPLINARY SOCIAL NETWORK RESEARCH

Chapter 1

Methods for Interdisciplinary Social Network Studies

Xiaoming Fu, Jar-Der Luo, and Margarete Boos

Contents

1.1 Introduction

1.2 Methodology for Combining Big Data Mining and Qualitative Studies in Theory Building

1.3 A Tour of Interdisciplinary Approaches and Case Studies Presented in this Book

1.3.1 Part I: Methodologies for Interdisciplinary Social Network Research

1.3.2 Part II: Social Network Structure

1.3.3 Part III: Social Network Behaviors

1.3.4 Part IV: Social Networks as Complex Systems and Their Applications

1.3.5 Part V: Collaboration and Information Dissemination in Social Networks

References

1.1 Introduction

People participate in social networks when they interact with their families, friends, colleagues, and other individuals or groups. Social networks link people together via a common interest and/or other kinds of interdependencies. Today, the dynamics of social networks are often fueled by access to modern online platforms and high geographic/spatial mobility, resulting in greater interpersonal interaction. For example, Facebook, the most widely used online social networking service as of this writing, reported 1.79 billion (including 1.66 billion mobile) monthly active users as of September 30, 2016 (Facebook, n.d.). China’s Tencent, one of the largest Internet companies in the world whose subsidiaries provide, among other services, instant messaging (Tencent QQ) and the mobile chat service WeChat, reported 1.1 billion registered WeChat users as of January 22, 2015, and 570 million daily active WeChat users as of November 5, 2015 (DMR, n.d.). Social networks—whether they be online or real world—are of vital importance to modern societies in that they influence daily work, contacts, and leisure activities. Social networks enable interactions for collaborating, learning, and information dissemination within physical (i.e., real world) or virtual (e.g., online) social networks.

A social network is composed of individual nodes (persons, teams, or organizations) and the ties (also called relationships, connections, edges, or links) between these individual nodes. Together these form a graph-based structure that is often complex (see e.g., Barabasi, 2003). Given the widespread presence of online social networks and also real-world networks, it is interesting to understand how a tie is created; how the network functions; what its structure looks like; and how it evolves, stabilizes, adapts, and changes. For practical cases and applications, we need to know how these features can be leveraged, such as how to bring together the strengths of diverse technical or scientific disciplines in creative collaboration, to make business or political decisions, and to develop risk-reducing measures to mitigate or control risk, for instance, in epidemics or stock markets, or even to curtail rumors/spam. This book intends to present new methods and techniques that are synthesized from different research disciplines involved in the formation, analysis, and modeling of various social networks as well as their applications.

Most existing studies on social networks (e.g., Milgram, 1967; Freeman, 2004) either study the network as a whole regarding its structure with specific relationships in the defined population, or the network from an individual perspective (so-called egocentered networks). Many have also studied the consequences for individuals who are embedded in social relations and networks, focusing, for example, on the effects in terms of receiving social support or finding a job (e.g., Granovetter, 1973). Physicists; social, behavioral, and epidemic researchers; and practitioners have developed and collected a large body of hypotheses, models, and empirical findings on the structure, processes, and consequences of social networks, both real word and online. In the last decade, online social networks have gained particular importance in everyday life due to their facilitation of the intercommunication (i.e., social networking) among a rising share of the population in modern societies. Indeed, the new forms of online social networks open up vast opportunities for studying social networks. Most networks that were studied in the social science domain were targeted at small groups, due to financial and practical limitations in accessing the data (Gjoka et al., 2010). Barriers that once made physical social networks inaccessible have now been overcome as a result of the emergence of big data storage, processing and traffic-managing capacities, and numerous social media and other online platforms. However, existing work among the so-called nodes of social networks—persons, teams, and organizations—does not yet take full advantage of the opportunities provided through interdisciplinary studies, which remains generally confined to specific fields. The result is a more intra- than interdisciplinary focus with limited advances. Interdisciplinary cooperation between social, behavioral, and epidemiological research, on one hand, and physics and computer science, on the other hand, holds the promise of enormous advances in the analysis of the potential of online social networks, and that of large-scale social networks in general.

We are pleased to witness a handful of researchers working with people from different disciplines, developing and employing various methodical approaches for studying complex social networks. A subset of such efforts is included in this book. These projects have been carried out in the form of close interdisciplinary collaborations by researchers with backgrounds in complex systems, statistics, and computer sciences, together with medical, management, behavioral, and social sciences, who continue to develop methods for data mining, network analysis, theory building, and more generally the interdisciplinary social network analysis methodologies.

By interlinking the expertise from divergent disciplines, new results and considerable progress are achievable in social network studies, as evidenced by the results reported in this book. Although a small set of chapters were written by scientists from the same discipline, knowledge and experiences from other disciplines were adopted and exploited in these chapters, constituting a broader sense of hybrid intra- and interdisciplinarity.

1.2 Methodology for Combining Big Data Mining and Qualitative Studies in Theory Building

This section will begin with a methodology developed during several case studies (e.g., see Chapters 4, 5, 6 and 7). In short, this methodology starts with quantitative studies, mining sample data with selected hypotheses (based on preliminary knowledge gained from a literature review), followed by qualitative analysis (e.g., through sociological interviews and questionnaires) towards ground truthing; based on this, predictions about certain network properties, patterns, or indicators can be made. By iterating this process, which integrates qualitative and quantitative studies, several times, hypotheses can be tested and new models may be established or existing models refined.

Before going into details about the methodology, we briefly explain several terms that are frequently used in this book:

• Big data: data collected from the online world or other digitalized sources that are too complex or of a too huge volume to be analyzed by traditional data processing tools
• Small data: structured data collected from quantitative surveys performed in the real world or extracted from big data
• Complex system: a system consisting of elements plus the interactions between these elements
• Data mining: the process of finding predictors for a social phenomenon with little or no guidance of theories; in other words, extracting potentially useful (but yet-to-be-empirically-validated) patterns from data sources, for example, databases, texts, the web, images, etc.
• Ground truth: level of accuracy of the training set reflecting or approximating the real world or population under investigation
• Ground truthing: the process of garnering sufficiently representative data that reflects/approximates the real case
• Hypothesis testing: the process of designing an empirical study apt to falsify a hypothesis derived from theory
• Machine learning: similar to how humans learn from past experience, a computer (i.e., machine) system learns from data that represent some “past experiences” of the applied domain
• Qualitative approach: includes typical sociological methods such as interviewing, field observations, open questions’ surveys, case studies, etc., which offer a way for hypothesis testing
• Quantitative approach: includes data mining and hypothesis testing based on structured and/or big data
• Real-world social networks: physical networks (e.g., families, teams, and organizations)
• Online and other virtual social networks: social networks that are media based (Internet, satellite, cell, Wi-Fi, computer, etc.)
• Supervised learning: method of labeling prior available example data (so-called training sets composed of observations, measurements, etc.) with predefined classes, which are used to train a model or algorithm to classify new data/instances into ones of the predefined classes
• Theoretical model: a theoretical mechanism that explains how explanatory variables influence the target social phenomenon
• Modeling: a process of developing a theoretical model for testing against quantitative data
• Theory developing/building: a process that begins with intuitions or interpretations (articulated as hypotheses), for example, on data mining results, then gives the reasoning behind the intuitions or interpretations, building a model based on said reasoning, defining the variables in the model, and collecting data from the real world to test the model in order to test the theory
• Survey: a method for collecting quantitative information about items in a population (Creswell, 2013)
• Interview: a conversation between two or more people where questions are asked by the interviewer to elicit facts or statements from the interviewee (Creswell, 2013)
• Sampling: selection of observations to acquire some knowledge of a statistical population (Creswell, 2013)
• Sampling bias: a bias in which a sample is collected in such a way that some members of the intended population are less likely to be included than others (Creswell, 2013)

The methodology of a research cycle in social network research often begins with mining of online data, with the expectation that some interesting social phenomena will be identified. We then interpret these findings by way of either a comparison with existing theories and/or by creating our own preliminary theory. Using preexisting theories and/or our own preliminary theory as a guide, various qualitative methods, such as interviews, field observations, open questions surveys, case studies, etc., can be used. Qualitative studies provide us with an understanding of ground truth, which can be used to test the findings and interpretations derived from data mining. Through the combination of ground truth, existing theories, and/or our preliminary theory, a base for theory building and hypothesis development is established. Then a model based on the operative theory is built in order to predict new facts, and more sets of data are collected for testing the theoretical model. Oftentimes, there are ground truths checked by surveys in the real world that do not jibe with our interpretation of the results of data mining, and/or further examination of initial qualitative studies reveals further observations not accessible through the findings and interpretations gained from the first-stage data mining. This will lead to a second run of data mining and qualitative studies. This process is illustrated in Figure 1.1.

The whole process of theory development concerning a social phenomenon includes several runs of data mining, interpretation, qualitative studies, and model building. Online big data opens up a new world for mining social science data upon which to build theories and for testing hypotheses to confirm theories. However, without checking the ground truth of online-mined data against real-world qualitative studies and quantitative surveys, the mining of online data remains invalidated and therefore largely useless.

Taking Chapter 7 as an example, where data about cooperation networks in the Chinese venture capital (VC) industry (based on the SiMuTon database) are explored, the authors try to understand the relational circle of leading companies in this industry. Just like a Dunbar circle (Dunbar, 1992), an industry leader has several layers of partners in his/her egocentered network, differentiated by the frequency of their cooperation. A high ...

Table of contents

• Cover
• Half Title
• Title Page
• Copyright Page
• Table of Contents
• Foreword
• Preface
• Editors
• Contributors
• Part I Methodologies for Interdisciplinary Social Network Research
• Part II Social Network Structure
• Part III Social Network Behaviors
• Part IV Social Networks as Complex Systems and their Applications
• Part V Collaboration and Information Dissemination in Social Networks
• Index