🔬 MOC — Research Pipeline

🔬 Research Pipeline Map of Content

Overview

This MOC maps the complete research methodology pipeline used in PUMA, combining three integrated frameworks.

---

Framework 1 — Evidence-Based Research Augmented by AI (EBSE + PRISMA-trAIce)

Phase 0: Problem Definition
    └── FINER criteria validation
    └── Research question formulation
    └── Falsifiable hypotheses (Popper)
          ↓
Phase 1: Systematic Literature Review (SLR)
    └── PRISMA 2020 protocol
    └── PRISMA-DFLLM extension (AI-assisted screening)
    └── PRISMA-trAIce (AI use documentation)
          ↓
Phase 2: Gap Analysis
    └── Reproducibility gap
    └── Prompting strategy gap
    └── Carbon footprint gap
          ↓
Phase 3: Methodological Design (DSR)
    └── Artefact specification (SDD/BDD)
    └── Experiment design
          ↓
Phase 4: Experimentation
    └── Controlled conditions
    └── Statistical analysis (Wilcoxon)
    └── Carbon measurement (CodeCarbon)
          ↓
Phase 5: Evaluation & Discussion
    └── H1, H2 testing
    └── Validity analysis
    └── Replicability verification

Key notes:

• PN-DSR-SLR-Methods — SLR + PRISMA protocol
• PN-Wilcoxon-FINER-Cornell-PRISMA — FINER + PRISMA details
• PRISMA-Log — Active screening log
• PR-PUMA-Ch2-Ch3-Ch4-Ch5 — SLR execution

---

Framework 2 — Design Science Research + SDD/BDD + AI Agents

DSR Cycle (Hevner 2004 + Peffers 2007):
    Problem → Objectives → Design & Development → Demonstration → Evaluation → Communication

SDD Chain:
    Spec-First Design
        └── OpenSpec format
        └── Spec Kit validation
        └── BDD scenarios (Given/When/Then)
        └── TDD test cases
              ↓
    AI-Assisted Generation
        └── Agent generates code/specs from prompts
        └── Human validates (Human-in-the-Loop)
        └── Red Teaming (rival hypotheses)
              ↓
    BMAD Workflow
        └── Brainstorm → Map → Architect → Develop

Key notes:

• PN-DSR-SLR-Methods — DSR paradigm + SLR
• PN-SDD-Framework — Spec-Driven Development + BDD + BMAD
• SP-Architecture — PUMA architecture
• SP-PUMA-Constitution — Non-negotiable principles

---

Framework 3 — MIT Student Method as Cognitive Operator

The MIT Student Method operates at three levels in PUMA:

Level	Application	Tools
Reading/Synthesis	Cornell notes + DRCA on each paper	PN-MIT-Student-Method
AI Interaction	RCOIF/RCOF + EGI + AMI + IIPR patterns	PN-RCOIF-Framework
Red Teaming	Rival hypotheses, counter-evidence search	PN-KeyConcepts-Agents-Reproducibility-RedTeam

Key notes:

• PN-MIT-Student-Method — Summary method (Cornell + Red Teaming)
• PN-MIT-Student-Method-Complete — Complete Q1/Q2/Q3 + Keshav integration
• SRC-MITAILab-WP316 — Original MIT AI Lab WP316 source
• SRC-Keshav-2007-HowToReadPaper — Keshav 2007 source
• PN-EGI-Framework — Guided Interactive Exploration
• PN-AMI-DRCA-IIPR-Frameworks — AMI + DRCA + IIPR
• PN-KeyConcepts-Agents-Reproducibility-RedTeam — Red Teaming practice
• WF-MIT-AILab-Reading-Workflow — Reading workflow
• WF-Research-Experiment-Routines — Experiment routines

---

Phase-to-Framework Mapping

PUMA Phase	Primary Framework	MIT Student Method Role
F0 Initiation	EBSE/SLR	Panoramic mapping + screening
F1 Design	DSR + SDD	Cornell on requirements → BDD stories
F2 Prototype	DSR + BDD/TDD	CoT + Few-Shot for spec generation
F3 Extension	DSR + BMAD	IIPR + AMI for agent iteration
F4 Analysis	EBSE + Stats	DRCA on results
F5 Closure	Full integration	Red teaming conclusions

---

AI Use Documentation (per PRISMA-trAIce)

TABLE tool, phase, purpose, validation
FROM "50 - Areas/51 Research"
WHERE type = "ai-use-log"
SORT phase ASC