v0.7.0Reliability, Discovery & Reading
Current release
Reliable public research and evidence-led reading
Made the public research system resilient to upstream database failure and reorganized reading around claims, evidence, limitations, and review state.
Included in this update
- Replaced per-request dynamic public research rendering with ISR and a versioned last-known-good snapshot.
- Added structured upstream error logs, a public research health check, and incident documentation for the unavailable Supabase project.
- Rebuilt the homepage around the active Discover → Evaluate → Claims → Evidence → Human Review → Framework loop.
- Added stable bilingual Claim pages, evidence groupings, public revision history, related papers, topics, and framework links.
- Redesigned paper reading around findings, study design, evidence strength, limitations, non-claims, HIE interpretation, and original sources.
- Corrected locale HTML language, canonical and hreflang metadata, structured data, share metadata, and content-based sitemap dates.
- Added paper search and topic, year, evidence-level, and human-review filters.
v0.6.0Phase 3
Evidence integrity and human review
Separated scientific evidence, AI assessment confidence, source scope, and human decisions while preparing a protected review workflow.
Included in this update
- Added explicit metadata, abstract, full-text availability, full-text AI review, and human-review levels.
- Added source scope, provenance reference, model, prompt version, and input hash fields to every evidence relationship.
- Added append-only revision histories for claims and evidence relationships without deleting prior research states.
- Protected the review inbox with Supabase Auth and a separate reviewer allow-list, with Approve, Edit, and Reject actions.
- Updated public paper and claim labels so AI-reviewed scope, scientific strength, AI confidence, and human status are visible separately.
v0.5.0Phase 3
Public evidence transparency
Turned the paper catalog into a bilingual, inspectable evidence library and published the first claim-to-paper relationships.
Included in this update
- Added English and Chinese detail pages for all 13 public papers.
- Published AI-authored summaries, evidence strengths, limitations, scores, confidence, and source-verification dates.
- Released four normalized claims connected to 12 directional evidence relationships.
- Kept raw model responses and internal research runs private behind RLS while exposing only a narrow public assessment layer.
v0.4.2Phase 3
Cognitive offloading evidence collection
Added a second AI-screened collection on external memory, digital attention, metacognitive calibration, and cognitive offloading.
Included in this update
- Verified and evaluated seven papers using publisher, PubMed, PMC, and Crossref metadata.
- Paired the influential smartphone brain-drain experiments with a later meta-analysis that qualifies the effect.
- Expanded the public catalog to 13 papers with complete authorship, topic, evaluation, limitation, and provenance records.
- Replaced the outdated Phase 1 ingestion note with the active evidence-screening method.
v0.4.1Phase 3
First public research collection
Published HIE Lab's first AI-screened research collection on human-AI complementarity, productivity, learning, and judgement.
Included in this update
- Verified six papers against publisher pages and Crossref metadata.
- Recorded authorship, topic assignments, AI evaluation scores, limitations, and provenance in Supabase.
- Published only papers that passed the relevance, evidence-quality, methodological-rigor, and confidence thresholds.
- Added the first live research records to the English and Chinese Papers pages.
v0.4.0Phase 3
Automated research pipeline foundation
Established a zero-manual-entry paper discovery and AI evaluation loop with conservative publication gates and a complete audit trail.
Included in this update
- Added a daily OpenAlex discovery job with rotating HIE research queries and bounded candidate volume.
- Added strict OpenAI structured evaluations for relevance, evidence quality, methodological rigor, novelty, and confidence.
- Added immutable evaluation records and idempotent research-run history in Supabase, private behind RLS.
- Added deterministic publication thresholds so low-confidence or metadata-only records remain internal.
v0.3.0Phase 2
Research data foundation
Established the migration-first Supabase foundation and added a permanent, bilingual record of product changes.
Included in this update
- Modeled papers, authors, citations, topics, claims, and evidence relationships in PostgreSQL.
- Added private-by-default RLS policies, topic seed data, schema checks, and typed Supabase clients.
- Connected the Papers route to a safe public-read state without exposing internal research data.
- Added bilingual Updates pages and centralized footer version metadata.
v0.2.0Phase 1
Bilingual publishing foundation
Expanded the original site shell into a complete English and Simplified Chinese experience.
Included in this update
- Added Chinese routes under /zh for every public section.
- Added an English and Chinese language switch across the site.
- Localized page copy, metadata, language alternates, and sitemap entries.
- Added visible release version and update date information to the footer.
v0.1.0Phase 1
Project foundation
Launched the first deployable HIE Lab website foundation with a restrained research-institute visual system.
Included in this update
- Created the Next.js and TypeScript application foundation for Vercel.
- Added the home, research, papers, claims, framework, journal, lab, and admin routes.
- Established the HIE Lab identity, typography, spacing, and evidence-first editorial direction.
- Prepared environment configuration, GitHub deployment, and project documentation.