CBGP-Databases

An ontology-driven, metaprogrammed database platform for running a research institute’s administration — projects, members, and publications — built so that the people who curate its data model day to day never need to touch code.

Contents

• Philosophy & Design
  • The one-sentence version
  • Why this exists
  • How it actually works, without the code
  • Strict separation of code from structure
  • Multilingual by the same mechanism, not as an afterthought
  • Is this novel? Where it sits relative to other approaches
  • A reusable pattern, not a one-off
  • Further reading
• Data Model
  • Two layers, one record
  • How values are typed
  • dcterms:type: which form wrote this record
  • Crossing both layers in one query
  • How this connects to the ontology
  • Interface text lives in the ontology
  • Worked example: is a field required, defaulted, or calculated on this form?
  • Worked example: one record that points at two others
• Installation
  • The two-store database
  • Running via Docker
  • Choosing a version
• Configuration
  • Login accounts
  • Ontology
  • Virtuoso connection
  • Admin email notifications
  • Activity triggers
  • AI agent access (MCP)
  • Verbose SPARQL/debug logging
  • A note on security
  • After changing .env
• Backup & Migration
  • What actually needs backing up
  • Making a backup
  • Scheduling automatic backups
  • Restoring from a backup (disaster recovery)
  • Migrating to a new server
  • One-off data fixes
• Admin Guide
  • Data Entry
  • History & Snapshots
  • Search & Queries
  • Cross-References
  • Exports
  • Asking questions through an AI agent
• User Guide
  • Logging in
  • Submitting a form
  • Announcing a project application
  • What happens after submitting
• Time Travel
  • What this is
  • Two kinds of question
  • Worked example 1: a single record’s full history
  • Worked example 2: a question about a period of time
  • Output format
  • Discovering what fields exist
  • Try it yourself