Documentation
¶
Overview ¶
Command validation is the M1 validation run: ARCHITECTURE_PHASE1.md §10's "20-paper / 5-field" check, run against live OpenAlex before v0.1.
Each seed gets a library of its own, so every number below belongs to one seed and nothing else. For each: add it, expand once with the default budget (500) and depth (3), count probable title duplicates (D14), and export GraphML — then check the export reads back with every edge resolving to a node in the file. One HTTP cache is shared, so a re-run is nearly free.
Seeds in medicine, physics, computer science and the social sciences are hand-picked: well-known papers a researcher would plausibly start from. The four humanities seeds are chosen the way spike 4 chose them — the most-cited arts-and-humanities articles with more than 20 references — because humanities DOIs are the ones least safely recalled. Two of OpenAlex's top "Arts and Humanities" results (a 1949 population-genetics paper, a sociology classic filed under Music) were skipped as misclassified.
No contact email is sent.
Usage:
go run ./spikes/8-validation > docs/validation-run.md