read-only

Query the graph

The filters tools/query/kg_query.py takes, as a web page. They combine with AND, exactly as the tool combines them, and each one reads a field the graph already stores rather than computing a judgement of its own. The closure toggle adds the rows the active declarations license and marks every one of them as derived.

Nothing on this page writes anything. The panel reads the published dataset at /graph.json and the declarations beside it; the same query on the command line prints the same rows with a footer naming the graph it ran against. Two results carrying different seed digests are not comparable, which is why the digest sits under the results rather than in a help page.

assigned, or one of the three ways an edge carries no family
node id, else substring of an id or a label
same matching rule as subject
a reference cites this CW volume
read off the reference, not the edge
substring of relation_family.by

Reading the dataset…

receiptsgraph state: 589 nodes / 2091 edges; families 1669 assigned, 354 pending-adjudication, 65 no-family, 3 text-undecided; seed.json sha256:9dc668bfff36; declarations relation_declarations.json from FAMILY_DEFINITIONS v0.7.1, 7 active