🌐 What is Tri-Covenant Watch?
Three independent NGO platforms — each focused on one UN human rights covenant — share a common evidence pipeline and cross-reference each other's data. Each platform is hosted and edited independently, but the cross-platform analytical pages (Tri-Covenant Dashboard, structural findings, executive summary, etc.) are co-produced and synced byte-for-byte across all three platforms.
The three platforms
♀ CEDAW (Women)
Convention on the Elimination of All Forms of Discrimination Against Women.
Host: Taiwan National Mothers' Alliance for Family and Children (台灣全國媽媽護家護兒聯盟).
Stance: opposes interpretation expansion of CEDAW beyond biological sex protection. Tracks "framework expansion vs women's substantive rights" (Axis 4 of the platform).
☻ CRC (Children)
Convention on the Rights of the Child.
Host: Alliance for AABE (國教行動聯盟). Education-policy advocacy NGO.
Tracks state failures in child protection, education, and emerging adolescent rights issues including online safety, mental health, and adolescent SOGIESC tensions.
⚙ CRPD (Disability)
Convention on the Rights of Persons with Disabilities.
Host: Cross-NGO collaboration (DPO + ally NGOs).
Tracks systemic implementation gaps including reasonable accommodation paradoxes, disabled women's intersectional rights, and indigenous disabled persons.
🔧 Methodology
The platform uses a reproducible four-step pipeline applied to all three covenants:
pdftotext (poppler), with fallback to Tesseract OCR (chi_tra+eng @ 300 DPI) for scanned PDFs.passage, policy_issue, rel_issue_passage, document, etc. Full-text search uses SQLite FTS5. Total: 102,728 segments across 3 platforms (as of 2026-05).PI Evidence Scoring Formula
Each Policy Issue (PI) receives a 0-100 score based on five dimensions:
- Passages (0-40) — number of government-document passages linked to this PI (50 passages = full score)
- Events (0-20) — advocacy events tracked (10 events = full score)
- Co-signatories (0-15) — number of cross-NGO co-signed positions (10 = full score)
- Distinct organizations (0-15) — diversity of issuing government bodies (5 = full score)
- Year span (0-10) — temporal coverage breadth (~14 years = full score)
Grade thresholds: A ≥ 70 (advocacy-ready) · B ≥ 50 (refine + advocate) · C ≥ 30 (focused supplementation) · D ≥ 10 (foundational research) · F < 10 (FOI / fieldwork required).
📊 Open Data
- Per-platform DBs: SQLite files in each platform's
data/directory. Schema documented in each platform's repository. - Cross-platform JSON outputs:
data/cross_platform/on the CEDAW platform —monthly_snapshots.json,keyword_comparison.json,crpd_coverage.json, etc. - Cross-platform search index:
api/cross_platform_search.json(1,658 indexed entries). - Search UI: cross-platform-search-en.html with FTS5-backed live search.
📜 Citation
Suggested citation:
Tri-Covenant Watch (Taiwan). (2026). Cross-covenant monitoring of Taiwan government discourse on CEDAW / CRC / CRPD: Structural findings and PI evidence scorecard. Available at: cedaw.taiwanmommies.org · naer-tw.github.io/child-rights-watch · naer-tw.github.io/disability-rights-watch. CC BY 4.0.
For specific findings, please cite the individual PI brief or structural finding URL.
📋 License
All cross-platform analytical content (HTML pages, JSON exports, methodology documentation) is released under Creative Commons Attribution 4.0 International (CC BY 4.0) — free to share and adapt with attribution.
The shared evidence pipeline (Python code) is open source under MIT License at github.com/naer-tw/evidence-pipeline. For replication, see the Developer / Replication Guide — step-by-step setup for a Tri-Covenant-Watch-style platform on any UN treaty.
Platform-specific NGO advocacy positions, however, remain the editorial property of the hosting NGOs and represent their respective stances, not necessarily those of the other platforms.
🤝 Cross-Platform Collaboration Model
The three NGOs share technical infrastructure (DB schema, build pipeline, cross-platform analytical pages) but maintain editorial independence:
- Each platform's
policy_issueentries reflect that platform's NGO's framing - Cross-platform pages (this one, structural findings, intersectional topics, etc.) present pure quantitative comparisons without editorial position
- Disagreements between platforms — e.g. on SOGIESC framework expansion — are themselves visible as data: see CEDAW Axis 4 vs CRC Axis 4 vs CRPD A14 framing
This model demonstrates that NGOs with distinct (sometimes opposing) advocacy positions can co-produce neutral cross-cutting evidence infrastructure, surfacing shared blind spots without forcing position alignment.
📚 Audience-specific entry guide
Different audiences find different entry points. Five "30-second to key content" paths:
📰 Journalists
1. Read 📋 Executive Summary (1-page): 6 findings + 4 asks
2. Open 📰 Case studies: 6 deep-dives with key numbers, intl comparisons
3. Citation format: see "Suggested citation" in release-notes
🎓 Academic researchers
1. GitHub repo: full pipeline (MIT)
2. CHANGELOG.md: design rationale + decision log
3. 📋 Release Notes: citation-ready external reference
🌐 UN treaty body reviewers
1. 📋 Executive Summary EN: 1-page brief
2. 🔬 6 Structural Findings EN: visualization dashboard
3. 📄 Joint Brief Template: NGO joint advocacy template
🤝 Domestic NGO advocates
1. 📋 PI Evidence Scorecard: rank issues for advocacy
2. 🎯 D-grade Research TODO: 4-dim action plan for low-score issues
3. 📄 Advocacy Brief Template: A4 printable
🛠 Replicating organizations
1. DEVELOPER_GUIDE.md: 7-step bootstrap
2. AUTOMATION_README: monthly workflow
3. Fork repo → adapt to other treaties (CAT/ICCPR/ICESCR/CERD)
📅 Long-term tracking
1. RSS: feed.xml · Atom: atom.xml
2. 📅 Monthly Tracker: 6 indicators baseline → target
3. 📈 Monthly Trend: total segments + Wave milestones
📞 Contact
For platform-specific inquiries, contact the hosting NGO directly via each platform's "About" page. For cross-platform technical questions, see the evidence-pipeline GitHub repository.