What this category's data looks like
Journal three-review-three-proofreading and editing quality materials include three-review forms, proofs, proofreading records, and receipts. The sources are the original documents from each stage of the editorial process, updated according to the manuscript circulation rhythm of each issue. Three-review form fields include manuscript title, review round, date, and signature; proofs include edition, proofreading round, revision marks, and responsible proofreader; proofreading records include page number, error type, quantity, and handling result. When these materials enter the knowledge base retrieval and recall function, the three-review forms serve as the candidate source for review rounds and the responsibility chain, the proofs serve as the candidate source for version and edition differences, the proofreading records serve as the candidate source for error quantities and handling status, and the receipts serve as the candidate source for signature completeness. During retrieval, select semantic, full-text, or hybrid mode according to the material form, so that candidate snippets cover different fields, for subsequent nodes or humans to trace exceptions and form review conclusions accordingly.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate snippets and similarity scores (after mode, merge, rerank, filter, and limit truncation). The exception list and review conclusions are then completed by subsequent nodes or humans.
| Object | Check | Verification action |
|---|---|---|
| Three-review form | The review round, date, and signature must correspond item by item with the proofs and proofreading records; the title and quantity of the same record must not fail to match across the two materials | In the search test, retrieve once each with synonymous questions, record the number of candidates and scores, and check whether semantic and full-text candidates both appear under searchMode |
| Proof | The edition, proofreading round, and revision marks must be consistent with the proofreading records; when the version is not synchronized, the proof date and signature shall prevail | View the order of the first few items once each with rerank enabled and disabled, use usingReRank to compare ranking changes, and check whether key passages are retained among candidates after similarity filtering |
| Proofreading record | The error type, quantity, and handling result must be supported by original records; when signatures or receipts are missing, it must not directly enter the review conclusion | Retrieve by page number and error type, check whether candidate snippets cover the quantity and handling result fields, record the number of candidates, and separately check the tokens of the quoted text and limit truncation |
| Receipt | The receipt signature and date must correspond with the three-review form and proof; when the date of the same record fails to match across the two materials, first check hits and associations | Retrieve with the receipt date and responsible proofreader, check whether candidates hit the corresponding review round and edition, and when the source fails to match, first check hits and associations rather than attributing it to quoted-limit truncation |
Three common errors
- Symptom: The title, quantity, or date of the same manuscript fails to match between the three-review form and the proof. Cause: The two materials are filled in at different stages, the fields are not aligned, and after retrieval hits, they are not verified item by item.
- Symptom: The proof edition and the proofreading record version are not synchronized, and candidate snippets point to the old version. Cause: The update rhythms are inconsistent, and retrieval is not filtered by date and edition, causing old materials to be mixed into the candidates.
- Symptom: The review conclusion lacks signature or receipt support. Cause: When the materials enter retrieval, the signature field is not retained, the candidate snippets do not cover the receipt, and subsequent nodes or humans directly generate conclusions based on them.
How to verify the setup
Feature acceptance: test once each with synonymously rewritten questions, compare candidate snippets and similarity scores, then toggle rerank to view the ranking of the first few items, and use searchMode to switch among semantic, full-text, and hybrid modes to compare the number of candidates.
Business test: create an exception record, see whether the source material can be traced, and check whether the fields, dates, and signatures of the three-review form, proof, and proofreading record correspond.
- In the search test, retrieve once each with synonymous questions, record the number of candidates and scores, switch
searchModeto compare semantic and full-text candidates, and enable and disable rerank to view the order of the first few items. - Retrieve by page number, error type, and date, check whether candidate snippets cover the quantity and handling result fields, separately record the number of items in the quoted array, and check the tokens of the quoted text and
limittruncation.