{"$id":"https://vlsi.kr/datasets/signal/schema/0.1.0","$schema":"https://json-schema.org/draft/2020-12/schema","additionalProperties":false,"description":"One normalized signal record in the Korea Semiconductor Signal Dataset. Every record is an event or snapshot produced by an existing VLSI Korea collection pipeline, normalized to a common envelope with mandatory provenance.","properties":{"confidence":{"description":"verified = read from primary source; reported = credible secondary report, not officially confirmed; estimated = sampled/modeled directional value (methodology must be stated in payload.note).","enum":["verified","reported","estimated"],"type":"string"},"domain":{"description":"Signal domain (see SCHEMA.md §2).","enum":["hiring","patent","earnings","investor_disclosure","market","attention","prediction"],"type":"string"},"entity":{"additionalProperties":false,"properties":{"name":{"description":"Canonical English name (e.g. 'SK hynix', 'Samsung Electronics').","type":"string"},"ticker":{"description":"Exchange-qualified ticker where applicable, e.g. '005930.KS', 'NVDA'.","type":"string"},"type":{"enum":["company","person","market","index","fund","institution"],"type":"string"}},"required":["name","type"],"type":"object"},"event_date":{"description":"Date the underlying real-world event occurred (trade date, publication date, bar date). Never the crawl date.","format":"date","type":"string"},"event_type":{"description":"Domain-specific event type, snake_case. Enumerated per domain in SCHEMA.md §3.","type":"string"},"labels":{"description":"Controlled-vocabulary labels: theme:<slug>, segment:<slug>, product:<slug>, signal:<slug>, channel:<slug>, exchange:<code>, status:<slug>.","items":{"type":"string"},"type":"array"},"license_note":{"type":"string"},"observed_at":{"description":"ISO 8601 date or datetime when VLSI observed/captured the event. event_date vs observed_at preserves disclosure lag as data.","type":"string"},"payload":{"description":"Domain-specific facts. All numeric values must be traceable to source.locator. Disclosed ranges stay ranges (arrays of [min,max]); never collapse to a point estimate.","type":"object"},"record_id":{"description":"Stable, unique, human-readable ID. Format: sig-<domain-token>-<slug>. Never reused, never renumbered.","pattern":"^sig-[a-z0-9][a-z0-9\\-]*$","type":"string"},"source":{"additionalProperties":false,"properties":{"locator":{"description":"Repo-relative path of the pipeline file, or public URL of the primary document.","type":"string"},"name":{"description":"Originating pipeline or primary source name.","type":"string"}},"required":["name","locator"],"type":"object"}},"required":["record_id","domain","event_type","entity","event_date","observed_at","payload","labels","source","confidence"],"title":"Korea Semiconductor Signal Record","type":"object"}