Data Model Reference¶
Last verified: 2026-07-26 against dev at f0388abb.
The authoritative logical ER design is docs/design/api-logical-design.md. This page summarizes the main entities.
| Entity | Meaning |
|---|---|
| Assessment | Top-level work unit containing documents and batches. |
| Document | Uploaded file and metadata belonging to an assessment. |
| Batch | Named set of documents to process. |
| Processor | Schema/extractor combination that defines how to extract data. |
| Batch run | Execution of a processor over a batch. |
| Schema | Structured field definition for extracted output. |
| Extractor | Worker implementation that extracts fields from documents. |
| Extracted item | Extracted result associated with a document and batch run. |
| Extracted field | Field/value/confidence item inside an extracted item. |
| Correction | User-provided correction to extracted data. |
| Document processing status | Per-document processing state in a batch run. |
Relationship Rules¶
- An assessment contains documents and batches.
- A batch contains document associations.
- A processor combines an extractor and schema.
- A batch run uses one processor.
- A document can yield many extracted items.
- Extracted items contain extracted fields and corrections.
- Processing status tracks a document inside a batch run.
Deletion Rules¶
The design doc states cascade behavior for assessments, documents, extracted items, and related records. Deletions are intended to be idempotent, returning success when the entity is already absent.