RERUM Collection Management (RCM)

A collection management and exhibition tool, allowing for the creation of ad hoc collections of distributed or even unknown resources.

This is not a tool for collections of things that you have. It is for things that you know. The typical workflow may look like this:

1. Find a record that is interesting;
2. Make it more interesting;
3. Share the result.

These results are most likely to be a collection of objects, but may be a single detail page. In some cases, the entity you are attempting to define may not exist on the Internet, so you will make a place for it.

Core Thesis

RCM is not primarily a repository. It is not primarily a CMS. It is a knowledge graph workbench where:

• Identity can be minimal.
• Descriptions are attributable assertions.
• Evidence is preserved and surfaced.
• Ambiguity and disagreement are first-class.
• Collections are dynamic and annotation-generated.
• Original resources can remain wherever they already live.
• RERUM provides a durable store for JSON-LD objects, annotations, and publication signals.
• Viewers and visualizers are modular clients over the same data graph.

Design Principles

1. JSON-LD Everywhere — All internal structures expressible as JSON-LD. Private application objects may exist for convenience but must round-trip into JSON-LD without semantic loss.
2. Non-Destructive Annotation — RCM describes resources by reference. Source objects are never mutated.
3. Virtual Collections by Preference — RCM prefers collecting remote resources over copying them.
4. Identity Through Evidence — A node may exist because there is evidence that something exists, even without authority records, images, or catalog records.
5. No Human Confidence Scores — Scholarly assertions carry evidence, attribution, date, motivation, and provenance. Numeric confidence is reserved for machine-generated guesses.
6. Chonky Nodes — Dense, annotation-rich identity anchors built from many small, attributable, evidence-bearing assertions.

Standards & Ecosystems

RCM is designed to interoperate with:

• JSON-LD — primary data format
• Web Annotation (W3C OA) — annotation model
• IIIF — image and manifest presentation
• Dublin Core — core metadata vocabulary
• FOAF — agent and social relationships
• GeoJSON-LD — spatial data
• CIDOC CRM — cultural heritage ontology
• schema.org — broader web compatibility
• Linked Data Notifications (LDN) — publication and announcement

Project Structure

/
├── _planning/          # Planning artifacts, concepts, examples
├── src/
│   ├── core/           # Core model: Thing, Representation, Expression, Annotation
│   ├── connectors/     # Resource resolvers and external API connectors
│   ├── storage/        # RERUM client and persistence layer
│   ├── graph/          # Graph query, traversal, and manipulation
│   ├── viewers/        # Modular viewers: IIIF, PDF, A/V, map, etc.
│   └── ui/             # User interface components
├── schemas/            # JSON-LD contexts and JSON Schema validation
├── examples/           # Sample JSON-LD objects and annotation bundles
├── tests/              # Test fixtures and test suites
├── docs/
│   ├── architecture.md
│   ├── data-model.md
│   ├── annotation-profiles.md
│   └── decisions/      # Architecture Decision Records
└── README.md

Development Phases

Phase	Status	Description
Phase 0	✅ Complete	Repository shaping, ADRs, documentation
Phase 1	✅ Complete	Core model, JSON-LD shapes, serialization, test fixtures
Phase 2	✅ Complete	RERUM client layer, LDN announcements
Phase 3	✅ Complete	Resource resolver / connector layer, graph engine
Phase 4	✅ Complete	Viewers (IIIF, PDF, Map, Network, Detail)
Phase 5	✅ Complete	Project profiles and configuration
Phase 6	✅ Complete	Entity birth, eventities, evidence classes
Phase 7	✅ Complete	Publication, export, static site generation

Planning

Detailed planning documents live in the _planning/ folder. See _planning/README.md for an overview.