What this category's data looks like
This page focuses on business-basis queries in the journal three-review-three-proof and editorial quality scenario. Material types include three-review forms, proofs, proofreading records, editing and processing records, and version receipts. Three-review forms carry the fields, dates, and signatures of initial review, re-review, and final review; proofs carry the layout proof stage, revision marks, and responsible proofreader signatures; proofreading records carry proofreading rounds, error registrations, quantities, and dates; version receipts carry issuance and archiving information. Materials are submitted by the editorial department and the proofreading department by issue or batch, and the update cadence rolls forward with proof stages and final-review issuance. After materials enter this feature, they first form visible data rows through parsing and chunking, then enter index flow; three-review forms, proofs, and proofreading records each participate in retrieval and back-referencing as independent material rows, used to locate the original-text basis when answering parents or writing it into a proposal.
Configuration and implementation checks
This step is executed by parsing and training tasks (server-side tasks), outputting data rows and index status (parsed, indexing, indexed). The basis entries that can back-reference the original text are then completed by retrieval and subsequent processing. After parsing is complete, parsing and chunking immediately persist to the database as visible data rows and bind a unique data identifier; the index flows through three states, data pending indexing does not participate in recall, and changing models only rebuilds data whose indexing is completed.
| Object | Check | Verification action |
|---|---|---|
| Three-review form | Fields, dates, and signatures are complete, and the same record name, quantity, and date can be matched against those in the proofs and proofreading records | Check the file records and data rows separately, open the data row to check status, and confirm it is in one of the three states parsed, indexing, or indexed; check the chunking method corresponding to trainingType and chunkSplitter |
| Proof | The layout proof stage, revision marks, and responsible proofreader signature are complete, the version is the current version, and unsynchronized old versions do not participate in back-referencing | Check whether the dates and signatures of the proof and the three-review form are consistent, and check whether the data rows covered by the rebuild task include only data whose indexing is completed |
| Proofreading record | Proofreading rounds, error registrations, quantities, and dates can back-reference the original text, and the entries can be matched to the original text | Check whether the names, quantities, and dates in the proofreading records and the proofs are consistent, and confirm that under the chunkSize setting the tokens and the number of array entries of the referenced text are recorded separately |
| Version receipt | Issuance and archiving information is complete, and versions lacking signatures or receipts do not enter the basis entries | Check the version numbers and dates of the receipt against the three-review form and the proof, and confirm that data pending indexing does not participate in recall |
Three common errors
- Symptom: the name, quantity, and date of the same record do not match between the three-review form and the proofreading record. Cause: the two materials are parsed and chunked separately, fields are not aligned across materials, and two sources appear during back-referencing.
- Symptom: the proof has been updated, but retrieval still returns the old version's content. Cause: the version is not synchronized, the old version's data row is still in the indexing-completed state, and the rebuild did not cover that entry.
- Symptom: a just-uploaded material cannot be found. Cause: the data row is in the pending-indexing state, the three-state index flow has not completed, and the retrieval layer naturally isolates it from recall.
How to verify the setup
This feature's acceptance: upload one material and observe the data row status flow and the coverage scope of the rebuild task; confirm the data row is in one of the three states parsed, indexing, or indexed and is consistent with the interface display, that the rebuild covers only data rows whose indexing is completed, that the initial training covers data pending indexing and data being indexed, and check the chunking and training mode corresponding to trainingType, chunkSize, and chunkSplitter.
Business test: ask a cross-material basis question, see whether the answer gives a back-referenceable source, and check whether the source can back-reference the original text of the three-review form, proof, or proofreading record.
- Open the data row to check status, confirm the data row is in one of the three states parsed, indexing, or indexed and is consistent with the interface display; check the file records and data rows separately, and confirm the identifiers of the file and the data row can be distinguished.
- View the data rows covered by the rebuild task, confirm it covers only data whose indexing is completed; check the fields, dates, and signatures in the three-review form, proof, and proofreading record, confirm the name, quantity, and date of the same record can be matched, and that versions lacking signatures or receipts do not enter the basis entries.