A Bio-inspired Classification System for Cyber-Physical-Human Identity Resolution
Abstract
The Internet has created a need for understanding complex technology and identities today.
Cybercrime can take years to solve, and a systematic design may aid in more rapid resolution.
Classification of identities involves the arrangement of shared qualities or characteristics
known as features set expression, useful for identifying specific types of cybercriminals
based on empirical evidence and logic. In order for classification of a cyber identity to be
explainable and acceptable to researchers, a holistic systematic approach is beneficial to
organize natural and synthetic features of living and non-living organisms into a standardized
model. Proven scientific methods in the biology domain are primary inspirations in this
research. I defined one such classification system that takes input evidence from
Time/context-Cyber-Physical-Human (TCPH) dimensions and infers generalized to specific
classes of individuals using an ontology with restrictive logic and bio-inspired visualizations.
A sharable and extensible classification system with a common vocabulary can be used as a
reference for understanding effective identity feature composition, expression, and logic.
What if a method exists that can help people understand with whom they are really
communicating on the Internet, what the intentions are, and if the other identity is human?
To answer the question, an approach needs to support emerging technology and human
behavior and should be codified into a systematic broad identity classification scheme with
built-in logic to study, experiment, and map related evidence over time. The problem area
currently involves disparate static tools and a lack of understanding of technology, humans,
and identities. Although several classification methods exist in various domains, an
extensible systematic design for linking a range of identities does not and would be beneficial in solving cybercrime cases. Today’s methods often produce inconsistent or unexplainable
results when applied to nondeterministic human behavior patterns. In addition, the changing
Internet infrastructure and emerging technology creates a gap that necessitates a solution that
consistently evaluates key identity attributes effectively over time. The solution could aid in
understanding combined types of identities temporally linked to multidimensional elements
for faster resolution. Challenges exist in creating an identity specification and designing a
modular classification method to support emerging features and accumulated evidence.
This research introduces a novel classification scheme consisting of feature sets and logic,
all mappable to a broad range of quantified identity classes such as humans, generalized
profiles, targeted cybercriminals, and unique person. The design was inspired by biological
vocabularies and visualizations for systematic, shared use such as the gene ontology and
genome goal for a complete mapping and understanding of all human genes. In this work,
TCPH aspects were designed within a classification system and evaluated with simulated
representations of real-world cybercrime and identity evidence. The design is instantiated
with prototyped ontology property axiom descriptive logic and simulated evidence to
demonstrate problem resolution effectiveness. The classification system was assessed with
multiple case trials to determine effectiveness and accuracy of extensible identity modeled
classes. Research contributions attempt to revolutionize the cyber-physical problem space
with a formal holistic specification of identities and prototyped design that classifies inferred
identities based on mapped multidimensional feature evidence expression logic. The phased
design approach introduces an initial concept in Chapter 4 that advanced into a redesigned
bio-inspired identity classification system prototype with four interconnected, trackable
TCPH dimensions in Chapter 5. Experiment trials and visualizations which demonstrate
theoretical salience and variance provide new insights for understanding and sharing
knowledge of present and future identities. Research goals include a classification system
experiment platform that supports a common identity understanding, general to specific class
resolution, and aid in comprehension of future complex hybrid identity composition.