International Journal of Recent Technology and Engineering (IJRTE)
ISSN: 2277-3878, Volume-8 Issue-3, September 2019
1278
Published By:
Blue Eyes Intelligence Engineering
& Sciences Publication
Retrieval Number: B2333078219/19©BEIESP
DOI:10.35940/ijrte.B2333.098319
Abstract: Social Networks are best represented as complex
interconnected graphs. Graph theory analysis can hence be used
for insight into various aspects of these complex social networks.
Privacy of such networks lately has been challenged and a
detailed analysis of such networks is required. This paper applies
key graph theory concepts to analyze such social networks.
Moreover, it also discusses applications and proposal of a novel
algorithm to analyze and gather key information from terrorist
social networks. Investigative Data Mining is used for this which
is defined as when Social Network Analysis (SNA) is applied to
Terrorist Networks to gather useful insights about the network..
Index Terms: Graph Theory, Graph Mining, Investigative Data
Mining, Social Network Analysis.
I. INTRODUCTION
A graph typically is a collection of non-linear nodes
connected using edges. A social network can typically be
defined using a graph, where the members of the network
comprise of the nodes and their mutual connections are
represented using the edges. Social network analysis is a key
technique in present day human communication analysis. It
has also gained a significant popularity in anthropology,
biology, communication studies, economics, geography,
information studies, organizational studies and human
psychology and has become a popular topic of speculation
and study. Humans have used the thought of social systems
freely for over a century to mean complex arrangements of
connections between individuals from social frameworks by
any stretch of the imagination scales, from interpersonal to
worldwide. Graph theory involves analysis of graphs and
extracting useful information and represents it using standard
metrics such as degree, etc.
Graph representation of SNA topologies are used for various
purposes such as community detection, network structures,
random walks and temporal networks. Social network
analysis as mentioned has been useful in numerous fields and
much of its impact is shared in [1] and [2]. There have been
numerous issues with these ever so important social networks.
A good compilation of these issues can be found [5] here.
Revised Manuscript Received on September 15, 2019
Krishna Ganeriwal*, SCOPE, VIT University, Vellore, India. Email:
ganeriwalk@gmail.com
Gayathri P, SCOPE, VIT University, Vellore, India. Email:
pgayathri@vit.ac.in
G. Gopichand, SCOPE, VIT University, Vellore, India. Email:
gopichand.g@vit.ac.in
H Santhi, SCOPE, VIT University, Vellore, India. Email:
hsanthi@vit.ac.in
In this paper, we discuss various metrics that can be applied to
social network graphs such as centrality, closeness,
betweenness, bridge, clustering coefficient, page-rank
centrality, etc. Furthermore, the existing algorithms and their
disadvantages are also analyzed and finally, a novel algorithm
is proposed to counter some of these disadvantages. In the
closing section of the paper potential areas for future research
is discussed.
II. GRAPH METRICS IN SNA
Typically social networks can be considered as unweighted
and undirected graphs for analysis and investigative purposes.
In these graphs the users represent nodes and the connection
between these users is illustrated using edges between the
concerned nodes. The resulting graph can be mathematically
formulated as an adjacency matrix (M
ij
) in the following
format:
M
ij =
This representation of the social network enables easy
calculation and representation of metrics and network
structure, mainly clustering coefficient, mixing time,
efficiency of the network, Eigen vectors and degree.
Following is an insight into these metrics and their use in
SNA:
A. Metrics
SNA Metrics is regarded as among one of the broadest
researches in graphical representation of Social network
analysis. By knowing relations between various nodes inside
a network, we can compute useful information about networks
and essentially they can reveal network structures. Having
metrics additionally implies we can gather essential
information about the system and utilize the information for a
few purposes.
1. Centrality:
Centrality gives key insight into the importance of a node in a
particular network and has been well explained by the work of
Freeman [3]. There are a few basic types of centrality
measures used, which are:
a. Degree centrality: Provides information on how well
connected a node is in a given network and is represented
using:
b. Betweenness: Provides
numerical data on the extent to
which a node intermediates the
Data Mining in Social Networks and its
Application in Counterterrorism
Krishna Ganeriwal, Gayathri P, G. Gopichand, H. Santhi