Purdue University · Doctor of Technology Candidate
Designing explainable, ontology-aligned, and evidence-traceable AI systems for cyber threat intelligence.
I am a cybersecurity researcher, doctoral candidate in Purdue University’s Doctor of Technology program, and senior industry practitioner with more than two decades of experience in threat intelligence, security operations, enterprise cyber defense, and responsible AI. My work sits at the intersection of artificial intelligence, cybersecurity, knowledge representation, and explainable cyber threat intelligence.
My research investigates how AI-driven systems can transform unstructured cyber threat intelligence reports, security alerts, and heterogeneous threat data into structured, ontology-aligned, and decision-ready intelligence. The work emphasizes explainability, evidence tracing, graded reasoning under uncertainty, and adversarial validation so that AI-supported threat analysis can move beyond fluent summaries toward defensible intelligence products.
Designing AI systems that explain how cyber threat intelligence conclusions are formed, what evidence supports them, and where uncertainty remains.
Using cybersecurity ontologies, MITRE ATT&CK, STIX/TAXII, and related standards to structure cyber threat entities, behaviors, relationships, and hypotheses.
Preserving links between AI-generated intelligence outputs and the underlying source evidence, including reports, alerts, indicators, techniques, and contextual artifacts.
Representing ambiguity, partial evidence, and analytic uncertainty through interpretable graded reasoning and fuzzy inference methods.
Exploring multi-agent AI architectures in which specialized agents extract, align, critique, validate, and synthesize cyber threat intelligence.
“Toward Explainable, Ontology-Aligned, and Evidence-Traceable AI for Cyber Threat Intelligence”
My doctoral dissertation focuses on the design and evaluation of an explainable cyber threat intelligence framework that transforms unstructured threat intelligence and security alert data into structured, ontology-aligned, and decision-ready intelligence.
The research addresses a critical gap in AI-enabled cybersecurity: many AI systems can summarize, classify, or prioritize security information, but they often fail to provide a transparent reasoning structure that explains why a conclusion was reached, what evidence supports it, how uncertainty was handled, and whether the conclusion aligns with established cybersecurity knowledge frameworks.
The proposed framework integrates large language models, cybersecurity ontologies, knowledge graphs, fuzzy reasoning, and adversarial validation agents. The goal is not simply to automate threat analysis, but to produce intelligence outputs that are explainable, auditable, and suitable for high-stakes cyber defense contexts.
The dissertation will emphasize artifact-based evaluation rather than human-subject experiments. Candidate evaluation dimensions include ontology alignment accuracy, evidence traceability completeness, structural consistency, reasoning transparency, and adversarial robustness against incomplete, conflicting, or deceptive inputs.
Mercer, D. (2026). Why AI Gets Things Wrong: Understanding the limits of artificial intelligence. Mercer AI Research.
A practical introduction to how modern AI systems generate outputs, why they fail, and what those limitations mean for professionals, leaders, and everyday users seeking to use AI responsibly.
Mercer, D. (2026). AI Governance Without the Theater. Mercer AI Research.
A practitioner-focused examination of enterprise AI governance that separates meaningful governance, accountability, and risk management from symbolic compliance activity.
Current academic writing focuses on explainable AI for cyber threat intelligence, ontology-aligned threat reasoning, fuzzy reasoning under uncertainty, responsible AI, and evidence-traceable AI systems. Peer-reviewed publications, invited talks, white papers, and conference materials will be added as they mature into formal outputs.
A dissertation-aligned research framework for transforming unstructured cyber threat intelligence and security alert data into explainable, ontology-aligned, evidence-traceable, and decision-ready intelligence.
A research system concept that integrates specialized agents for extraction, ontology alignment, hypothesis generation, adversarial validation, and explanation synthesis across cyber threat intelligence workflows.
A semantic reasoning initiative focused on aligning cyber threat intelligence entities, indicators, techniques, vulnerabilities, and evidence relationships to operational frameworks and structured knowledge models.
A hybrid reasoning project exploring how neural extraction, symbolic representations, ontologies, and knowledge graphs can support explainable and defensible cyber threat analysis.
A research effort focused on representing ambiguity, partial evidence, weak signals, and analytic uncertainty in cyber threat intelligence through interpretable fuzzy rule-based systems.
I welcome opportunities for research collaboration, speaking engagements, professional dialogue, and interdisciplinary work at the intersection of artificial intelligence and cybersecurity.