Page 117 - KDU INTERNATIONAL JOURNAL OF CRIMINAL JUSTICE

KDU International Journal of Criminal Justice (KDUIJCJ)
Volume I | Issue II| July 2024

for insider threat and espionage detection. A Bayesian network is a graphical
representation that depicts the causal relationships between different variables

involved in the threat detection process. In the context of insider threat and
espionage, variables can include user behaviors, access privileges, job roles,
communication patterns, and external threat intelligence, among others.

The network structure illustrates the dependencies between these variables,
indicating how changes in one variable can influence the probabilities of other

variables. The modeling process starts with defining the variables of interest and
their relationships based on expert knowledge, domain expertise, and historical

data. Each variable is represented as a node in the network, and the edges
between nodes represent the dependencies and causal relationships. Prior

probabilities are assigned to the variables based on available information,
representing the beliefs or assumptions about their initial states. Updating

probabilities is a fundamental aspect of Bayesian networks. As new evidence or
observations become available, the network is updated using Bayes' theorem to
revise the probabilities of the variables. This allows for the incorporation of real-

time data and the adjustment of beliefs based on the observed behaviours or
events. The updated probabilities reflect the current state of the variables and

can help identify potential insider threats or espionage activities.

Bayesian network modeling enables the detection of suspicious behaviours and

anomalies by assessing the joint probabilities of multiple variables. For example,
if an employee exhibits unusual access patterns, the network can evaluate the
conditional probabilities of related variables, such as communication patterns,

job roles, or previous incidents, to determine the overall risk level. By
considering multiple factors simultaneously, Bayesian networks offer a more

comprehensive and accurate assessment of the threat landscape. Moreover,
Bayesian networks can facilitate proactive risk mitigation by enabling "what-if"

scenario analysis. By modifying the probabilities of certain variables within the

106