# IL knowledge graph

Historical transcription of the information-literacy model documented in the **4.0 (2016)** and **5.1 (2018)** manuscripts. Current graph revision: **1.1.2, 25 September 2026**.

## Open

Extract the entire ZIP, then open **`index.html`** in a browser. Keep the files together. The page has no network dependencies, requires no installation and makes no external requests. `sources.html` is the reader-facing bilingual source record. It contains the historical model table, the 15-competency crosswalk, the original wording, editorial English translations of the selected French manuscript excerpts, the two manuscript references, and the annex images.

A read-only local preview server is also included. With Node.js installed, run `node scripts/serve.mjs` from this folder, then open the loopback address printed by the command. It listens only on 127.0.0.1. The package does not modify or deploy any website.

## Contents and navigation

The graph contains **221 nodes and 453 source-attributed navigation relations**. Six views separate the architecture from its documentary detail:

- **Structure:** the eight pillars, nonlinear PRI, and reflection after the search. The six first pillars occupy one plane. A faint dashed navigability mesh visualizes the manuscript’s global any-to-any navigation claim without converting it into fifteen canonical pairwise relations. A separate dashed feedback guide links post-PRI reflection to a later PRI iteration. The wire enclosure depicts the source verb “encadre”; none of these guides are historical ledger assertions.
- **Competencies:** all 15 synthesized competencies, attached to their source pillar. The `A` prefix denotes the annex’s competency numbering.
- **Crosswalk:** the 39 X marks in Annex 3. Empty cells and NILIS hatching are retained in the record and source table, not converted into inferred endorsements or negative findings.
- **Quoted indicators:** all 181 indicator occurrences in Annex 2, including repetitions. Filter by pillar to inspect smaller groups. They are attributed to the source models, not presented as 181 independently authored competencies.
- **Historical context:** selected definitions from the manuscript bodies. The 4.0 and 5.1 formulations remain separate. The 5.1 formulation concerning independent information-processing systems and truth value is preserved rather than retrospectively corrected.
- **All:** the complete provenance graph. Historical nodes and relations are separated into presentation layers: model/process/pillars, synthesized competencies, quoted indicator occurrences, source-model provenance, and manuscript context. The concept index is complete in every view.

Drag to rotate the actual XYZ scene; Shift-drag or right-drag pans it. Wheel or pinch zooms. Clicking a node opens its record; double-clicking or the sidebar’s focus control centres it. Keyboard arrow keys rotate, +/− zoom, and 0 resets. The Expand button enlarges the view; Escape closes it. All nodes can also be selected from the keyboard-accessible index.

French and English interfaces are available. English titles, descriptions and competency wording are editorial translations. Original quotations retain their original language. The original text is not replaced when the interface language changes.

## Numbering is historical, not normalized

The bodies of both manuscripts assign **7 to metacognition and 8 to ethics/law**. Annex 3 explicitly assigns **7 to ethics, responsibilities and law, and 8 to metacognition**. This is preserved as a **source numbering inconsistency**. The evidence does not establish two intentionally designed conventions. The selector exposes both readings without merging or renaming the underlying concepts. Competency codes such as `A7.1` always refer to Annex 3.

The accompanying review records the erroneous claim in the newer archive that the annex lacked explicit numerals. The original vector text and image show them. That claim was not copied into this transcription.

## Lexical fidelity

The source wording is authoritative for the transcription. Editorial text may explain presentation choices, but it must not strengthen the manuscript’s verbs or convert them into more demanding normative claims.

This is especially strict for **“L’éthique, les responsabilités et les lois”** and the competency **“La personne infolettrée comprend et tient compte des enjeux éthiques et légaux qui entourent l’usage de l’information.”** The graph does not translate that wording into “obey the law,” “comply with the law,” or a claim that legality creates ethical authority. A law can itself be ethically objectionable. The HTML entry points contain an adversarial-reviewer warning so later revisions do not silently introduce compliance language.

## Foundation and rendering

The supplied Epistemethics graph’s **generic schema, validator, projector and dependency-query module are reused byte for byte** in `foundation/`. Its embedded stylesheet is retained as `assets/reference.css`. The header, palette, typography, graph/sidebar layout and concept-index pattern follow that reference, with bounded responsive and interaction additions in `assets/il.css`.

The renderer is **not the original Three.js/CDN runtime**. It is a self-contained software 3D renderer: XYZ coordinates, a perspective camera, orbit/pan/zoom, depth ordering, node hit-testing and collision-managed labels, drawn on Canvas. This makes the delivered scene independent of remote libraries and WebGL availability. The rendering substitution changes transport and presentation, not the historical source model or canonical ledger foundation.

`data/il.source.json` is the curated source/transcription map. `data/il.ledger.json` contains the typed records and source-attributed navigation assertions. `data/il.projection.json` is the unchanged generic projector’s output; `data/il.presentation.json` is the reader-facing projection. The source map and ledger are not replaced by the picture. Every canonical displayed relation has `inference: none`. Presentation-only guides returned by `assets/layout.js` are deliberately absent from the ledger and are rendered with a dashed visual language.

The generic assertion field `modality: possible` marks the proposed navigational representation inherited from the supplied construction pattern. It is not a numeric confidence assessment of a manuscript quotation. No historical ChangeEvents are invented.

## Rebuild and inspect

Run `node scripts/build.mjs` to rebuild the canonical ledger, projection, browser data and source reader from the included transcription map. Run `node --test tests/historical.test.mjs` for the structural and historical-source checks. Run `node tests/browser_check.mjs` for the dependency-free Chromium interaction and screenshot harness, and `node tests/package_check.mjs` for local-link, build, foundation, ledger and packaging checks. The package was checked with Node.js 22.16.0 and system Chromium.

`qa/` includes the test outputs and rendered screenshots. `REVIEW.md` states the findings and the exact inspection method. In this container, browser policy blocked file and HTTP navigation; the visual tests therefore inlined the exact local stylesheet and script bytes into Chromium’s `about:blank`. Native file-mode navigation was not counted as a test pass. The graph itself uses classic local scripts, not fetch requests, modules or remote assets.

`verify_originals.py` compares the selected paragraphs and all seven annexes against the two exact source DOCX copies. It uses Python’s standard library; run it inside a virtual environment. For example: create a venv with `python -m venv .venv`, then use that venv’s Python executable to run `scripts/verify_originals.py --doc4 <4.0.docx> --doc5 <5.1.docx>`. The originals are not redistributed in this package; exact source-copy hashes are recorded in `SOURCES_AND_TRANSCRIPTION.md` and the data.

The older Python/Playwright browser harness remains included for environments that already have Playwright. The current delivered QA was regenerated with `tests/browser_check.mjs`, which drives Chromium through the DevTools Protocol and requires no third-party Node package. `CHROMIUM_EXECUTABLE` can select another browser binary. `tests/package_check.py` remains available for a full Draft 2020-12 JSON Schema pass when `jsonschema` is installed in a venv; the dependency-free delivered package check is `tests/package_check.mjs`.

## Attribution and scope

Historical model and manuscripts: **2016–2018**. Source-model wording remains attributed to SCONUL, ACRL, ANZIIL, NILIS, IFLA and CILIP as cited in the manuscripts. Graph transcription, English interface translations, and 3D geometry are editorial work dated September 23, 2026.

This is a transcription of the IL model and its selected conceptual context, not a transcription of every chapter, case study, bibliography entry or scrapbook item in the two manuscripts. No later Five-Vow, FRAME or Epistemethics framework is added to the historical eight-pillar model.