Published dataset

A machine-readable publication of the current authored mAtlas graph. It supports retrieval and deterministic local analysis; it does not provide a remote query service.

Dataset contract

A source-backed visual atlas connecting mathematics, physics, and fundamental chemistry from formal structures and physical law through chemical composition, bonding, reactivity, measurement, and radiochemical processes.

Content version
2.3.0
Schema version
2.1.0
License
Creative Commons Attribution-ShareAlike 4.0 International (CC BY-SA 4.0)
Stable manifest
/data/latest/manifest.json

Current downloads

The /data/latest/ URLs are replaced on each publication. The manifest reports the current content version and SHA-256 digests. Save the manifest and downloaded artifacts when reproducible use requires a retained snapshot.

ArtifactURLFormat, SHA-256, size
Canonical graph JSONcurrent URLapplication/json
56bd59ea35cdcd6acbdbbed0b3cf46d1eaa2186e2eed0aaf29794a3faf4d7f67
1,964,490 bytes
Graph JSON Schemacurrent URLapplication/schema+json
51e728559179842248e3ca54bb8334b1dbca46536a7dd31514350c1b0b3fe356
7,153 bytes
Stories and Views JSONcurrent URLapplication/json
ce8aa776bca406af86edba34e6e6dbea31b55da13ff1c24cb56f0b6c2ee38b57
60,502 bytes
Share-link codec JSONcurrent URLapplication/json
59031f96bc8702320ce6dbb4fa5fb0e5dd91b6e468f3c89ffaccac5e71589032
2,075 bytes
Content provenance JSONcurrent URLapplication/json
e356e5df60e958cde8887d1390a3de4af5cbf64c14e8f3f9495e1738fee697b1
1,622 bytes
Concepts and construction junctions NDJSONcurrent URLapplication/x-ndjson
009b4afea8f2f274f765ba59e9d9951cbf3efd923a5e7b3950c0ee57b1b005f9
1,316,715 bytes
Relations NDJSONcurrent URLapplication/x-ndjson
6386c91b6161b40651294c87d16b83afc015c27d6161ec7afdcc6fe56eb905b4
4,880,909 bytes
Sources NDJSONcurrent URLapplication/x-ndjson
7a5f266c1de6a1563f66dbdf3c48340045f44b554add9aa3b4170c3579c65d60
1,017,497 bytes
Domains NDJSONcurrent URLapplication/x-ndjson
b534002f86d06d78368aa753b1f4e1cd8390e97fc029d393ecafc40da34f2250
21,866 bytes
Fields NDJSONcurrent URLapplication/x-ndjson
2217188321a0ebbcd79b59165080da50a1c54184d78acd0bc0f50ffce1015d4e
2,091 bytes
Relation types NDJSONcurrent URLapplication/x-ndjson
5d000a58fe0188b31514dd6875d091bdf423489fcc60c4c22e4b09f2c86ac82a
23,108 bytes
mAtlas SQLite databasecurrent URLapplication/vnd.sqlite3
0306d83988d8775cd13a6ac54f43eb53b94e333221d76160cea5d20922cf952b
8,507,392 bytes
mAtlas RDF JSON-LD graphcurrent URLapplication/ld+json
2fcf896e95572a0953f2550f48585a716a70f382855550642c0bbc60725e0e47
4,722,269 bytes
mAtlas RDF Turtle graphcurrent URLtext/turtle
97127e58802066c826a3bbbc55ba1774cfd2ae899edfdb86883349dfd9c0f5fc
3,435,824 bytes

Small deterministic records

Browser agents and integrations can retrieve individual records without first parsing the entire graph.

Use with an AI system

  1. Start with the data manifest or llms.txt.
  2. Resolve names to canonical concept IDs before making graph claims. Direct relations remain authored source → target assertions even when a path traverses an edge backwards.
  3. Use the published SQLite file or zero-dependency Python library for paths, closures, and subgraphs. Use predecessor and prerequisite traversal exactly as the relation-type metadata defines them.
  4. For a cited claim, use the canonical concept page and, for a direct edge, its relation fragment; retain the linked external sources attached to that record.
  5. Do not infer unrecorded graph edges from general knowledge.