UNSW Monetise v2 — System Architecture
Research Commercialisation Engine
How a research paper travels from academic record to commercial intelligence
↖ Click any station for details
Stage 01
Ingest⚙
OpenAlex API
UNSW Primo REST
⚡ UNSW Filter
Reject non-UNSW
✓ Pass affiliation
Fetches & deduplicates papers from two sources. Merges by DOI, prefers Primo abstracts.
Stage 02
Evidence🔬
Citations Trends News Patents Preprints EuropePMC Wikipedia Funding
Sources coverage
~7/11
8 independent signal sources. Each stored with URL snapshot for auditability.
Stage 03
Scoring📊
D1
×15%
D2
×20%
D3
×15%
D4
×20%
D5
×15%
D6
×10%
—
CRI SCORE /100
Claude LLM scores 6 dimensions. Rules enforce D4 cap if no patent evidence found.
Stage 04
Gate 1🔒
All must pass to score ≥70
D2 ≥ 3  (TRL)
D3 ≥ 3  (Market)
D4≥4 or D5≥4
AWAITING
Hard gate. Fail = CRI capped at 69. Pass = eligible for business plan generation.
Stage 05
Enrichment🧠
Claude LLM
Commercialisation strategy
Industry partner suggestions
IP & funding pathways
Business plan draft
Risk assessment
LLM generates actionable commercialisation roadmap. Only runs for Gate-1 qualified papers.
Output
Portal📋
74
CRI SCORE
HIGH CONF
Gate 1 Passed
Business Plan
7/11 Sources
Evidence Trail
Final intelligence displayed in portal. Score, evidence citations, business plan, audit trail.
Active Agents
ingest_agent
Running · ~2 s/paper
marketer_agent
Running · 8 sources
scorer_agent
Running · D1–D6
enricher_agent
Running · Gate≥70
critic_agent
On-demand · Audit
Data flow
Amber particles = active processing
Gate passed
Gate failed / rejected
LLM analysis