E.A.T. Explore

Seven small static web apps onto the E.A.T. (Experiments in Art and Technology) data in the Semantic Lab Wikibase at base.semlab.io, plus the data-connections report they were designed from.

Every page queries the live SPARQL endpoint (https://query.semlab.io/sparql) when it opens. There is no build step and no server-side code.

Site (docs/)

Path	App	Data path
docs/index.html	Landing page
docs/report.html	Data-connections report with diagrams
docs/documents/	Paper Trail: archive documents float on screen; click one to expand its connections hop by hop (documents are the interface)	document → mentions · subject · creator · depicts → entities → further documents
docs/matchmaker/	The Matchmaking File: engineer cards, scans, proposed artists, engineer–artist graph	card → engineer ⇄ artist
docs/nine-evenings/	9 Evenings night by night: performances, crews with roles, equipment, documents, press	event ← performance → work · contributor [role] · uses
docs/reader/	Read the Archive: documents as passages with linked mentions; entity → every passage	document ← block → mentions; claim → reference block
docs/whos-who/	Who's Who: person profiles assembled from all four projects	person ↔ everything
docs/timeline/	Archive Against Press: dated Knowledge Graph items vs bibliography references	item → date
docs/map/	Where the Engineers Were: geocoded card addresses, Knowledge Graph places, places of publication	address, coordinates
docs/technology/	Technology as the Join: works' uses terms vs engineers' fields of work, with reconciliation suggestions	term
docs/shared/	Shared stylesheet and SPARQL client (sparql.js)

Publish on GitHub Pages

Deployment is done by GitHub Actions (.github/workflows/pages.yml): every push to main that touches docs/ uploads that folder as the Pages artifact and deploys it. No build step; the folder is served as-is.

One-time setup after the first push: in the repository Settings → Pages, set Source to GitHub Actions. The site then appears at https://semanticlab.github.io/eat-explore/. The workflow can also be run by hand from the Actions tab (Run workflow).

All links inside docs/ are relative, so the site works under the /eat-explore/ sub-path as well as at a domain root.

Run locally

python3 -m http.server 8000 --directory docs

then open http://localhost:8000/. ES modules need http, not file://.

Notes

• Browser requests go straight to the query service; it sends Access-Control-Allow-Origin: *. Queries are tagged #semlab-data-process in their first line. Results are cached per tab in sessionStorage.
• Roles (engineer, performing artist, production team member…) are read from statement qualifiers (pq:P262 object has role, pq:P117 subject has role); provenance from prov:wasDerivedFrom … pr:P26 reference block.
• The Rauschenberg Foundation re-uploaded its archive files into /sites/default/files/2026-10/; the dated paths stored in the Wikibase 404. shared/sparql.js rewrites them on the way in (139 of 142 resolve; foundation_urls.py checks the mapping and writes docs/documents/foundation.json). Updating the P200/P3 values in the Wikibase would make the rewrite unnecessary.
• The map joins live address strings to docs/map/geocoded.json, produced once by geocode.py from Nominatim. Re-run it if addresses change.
• External libraries (d3 7.9.0, Leaflet 1.9.4) and fonts load from cdnjs and Google Fonts.

Data and scripts (repository root)

• data/items.json — all 5,773 items in the four projects with labels, descriptions, project tags and direct claims.
• data/entities_full.json — full Wikibase JSON (qualifiers, references) for the 1,737 items carrying link properties.
• data/properties.json, data/external_items.json — property definitions and items linked from outside the four projects.
• q.py — run a SPARQL query from the shell with prefixes set. dump.py, fetch_entities.py, analyze.py, analyze2.py, geocode.py — reproduce the dumps and the numbers in the report.
• eat-data-connections.html — source of the report page.