Research Methodology
Design Science Research approach to evidence-first decision systems
Design Science Research Framework
AIJIM is built on Design Science Research (DSR) methodology, which combines rigorous empirical investigation with practical system development. The approach follows a cyclical problem identification → solution design → evaluation cycle.
Research Program Structure
The AIJIM research program comprises three interconnected papers that form a coherent scientific contribution:
Paper 1: AIJIM Reference Model
Status: Accepted for publication (Business & Information Systems Engineering, 2026)
Establishes the theoretical foundation: the five invariants (I1–I5) that govern evidence-first decision systems. Defines the Data → Verification → Reporting → Knowledge (D→V→R→K) chain and formalizes the gap between ideal and actual evidence evaluation practices.
Paper 2: Evidence Verification Framework (EVF) Survey
Status: Under review
Empirical evidence: systematic review of 140 studies across investigative journalism, fact-checking, and AI transparency. Quantifies the gaps: seeds missing in 91.4% of studies, confidence intervals absent in 97.1%, artifacts incomplete in 72.9%, judge configurations entirely undocumented (100%).
Paper 3: Protocol Paper (System Implementation)
Status: In progress; frozen baseline plus RFC-009 publication-protocol delta
Solution design and evaluation: specifies the AIJIM enforcement architecture that closes the identified gaps. Demonstrates how the five invariants can be operationalized in production systems with formal verification, runtime monitoring, immutable publications, and three orthogonal falsifiability axes: ECAM-X attribution validity, AEI-Delta explanation stability, and TCA human reliance calibration.
The Research Cycle
The three-paper structure implements the core DSR cycle:
- Problem Identification: Evidence evaluation in practice lacks reproducibility artifacts and formal verification mechanisms (Paper 2 quantifies this).
- Solution Design: The D→V→R→K chain and five invariants provide a principled architecture (Paper 1 theorizes this).
- Artifact Development: AIJIM is a working system that implements the protocol (Paper 3 specifies and evaluates this).
- Evaluation: Runtime KPIs and reproducibility metrics demonstrate that the system closes identified gaps and advances decision quality.
PhD Research Context
Institution: University of Technology Sydney (UTS), Faculty of Science
Funding: EXIST Research Program (Grant 03EGTTH025)
Supervisor Team: Specializes in evidence systems, investigative practices, and AI governance
Focus: Domain-agnostic evidence-first decision systems with journalism as proof case
Relevance to Practitioners
While rooted in academic research, AIJIM's DSR methodology ensures practical applicability:
- Reproducibility: Every decision is backed by complete seed data, configuration, and artifacts (solving the 91.4% seed gap).
- Verifiability: Formal invariants ensure that critical properties hold at runtime, not just in theory.
- Measurable Impact: I1–I5 KPIs track system performance against research objectives in real investigations.
- Domain Flexibility: The evidence-first pattern generalizes beyond journalism to any field requiring reproducible analysis.