What this category's data looks like
Materials for Q&A intake and missing-item follow-up in peer review organizations come from three types of records: review invitations, expert review comments, and author revision notes. A review invitation contains the manuscript title, invitation date, and receipt signature; expert review comments contain comment items, submission date, and signature; author revision notes contain item-by-item responses, revision date, and signature. The three types of materials use the intake record as the entry point. When entering the multi-turn dialogue and prompt stage, the review invitation serves as the source of question items, the expert review comments serve as the basis for judgment, and the author revision notes serve as the completed result. The update cadence rolls with intake batches, and the version is synchronized after each round of follow-up questions. Field characteristics center on title, quantity, date, and signature. The same record often has mismatched title, quantity, and date across two materials, unsynchronized versions, or missing signatures or receipts.
Configuration and implementation checks
This stage is executed by the memory turns and prompt configuration in the application settings, and outputs the referenced historical message range and answer constraints. The "intake record and completed question items" are then completed by the model.
| Object | Check | Verification action |
|---|---|---|
| Review invitation | The review invitation's title, invitation date, and receipt signature must be consistent with the intake record; a missing receipt is judged as a missing item | Check whether the first-round question in messages carries the manuscript title and invitation date, and compare the signature field |
| Expert review comments | The number of comment items and the submission date must be consistent with the intake record; an unsynchronized version is judged as a missing item | Check the number of comment items and dates round by round under chatId, and check whether an old version of comments is referenced |
| Author revision notes | The item-by-item responses, revision date, and signature of the revision notes must align with the aforementioned materials; a missing signature is judged as a missing item | Check whether the revision note fields in variables correspond one-to-one with the review comment items |
| Continuous follow-up | The second round can correctly carry forward the references of the first round; when key information is missing, ask a follow-up question first before answering | Run a trial with two consecutive rounds of questions, and observe whether the second question restates the manuscript title and date from the first question |
Three common errors
- Symptom: The same manuscript has mismatched title, quantity, and date between the review invitation and the expert review comments. Cause: The two materials were entered separately, and the dialogue began without field comparison.
- Symptom: The author revision notes reference an old version of the review comments, and the version is not synchronized. Cause: The intake record did not lock the comment version, and earlier historical messages were brought into the follow-up questions.
- Symptom: A receipt or signature is missing, but the intake record continues to ask follow-up questions. Cause: Missing-item identification does not cover signature and receipt fields, and the follow-up direction deviates from the material basis.
How to verify the setup
Acceptance for this feature: run two consecutive rounds of follow-up questions, see whether the second question carries forward the references of the first question, and check whether the historical message range in messages and chatId matches the memory turns.
Business test: ask a question that lacks key information, and see whether it asks a follow-up question first before answering.
- Check the current value of memory turns in the application settings, and confirm it matches the prior context range required by the question.
- Run a trial with two consecutive rounds of questions, compare the usage records of the two runs, and confirm that usage changes accordingly after adjusting the number of turns.