What this category's data looks like
Grade entry and credit recognition materials include three categories: transcripts, deferred examination applications, and make-up examination arrangements. Sources are academic affairs system exports and departmental paper archives, and the update cadence rolls forward by semester batch and deferred/make-up examination nodes. Transcripts contain course name, credits, grade, and instructor signature; deferred examination applications contain application date, reason, approval signature, and receipt; make-up examination arrangements contain make-up examination time, location, and invigilator signature. When these three categories of materials enter knowledge base retrieval and recall, transcripts serve as the main record providing course and credit fields, deferred examination applications serve as status change evidence providing date and approval chain, and make-up examination arrangements serve as the basis for follow-up actions providing time and location fields. The field characteristic is that the name, quantity, and date of the same course across different materials need cross-checking, and the completeness of signatures and receipts determines whether handover can proceed.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate chunks and similarity scores (after mode, merge, re-rank, filter, and upper-limit trimming). The handover-ready deployment status and operation records are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Transcript | Course name, credits, grade, instructor signature, and date must be aligned item by item with the deferred examination application and make-up examination arrangement | In the search test, switch searchMode to semantic, full-text, or hybrid, and compare whether the candidate chunks cover the course and credit fields |
| Deferred examination application | Application date, reason, approval signature, and receipt must be consistent with the timeline of the transcript and make-up examination arrangement | Turn usingReRank on and off, compare the order of the top few candidates, and check whether approval chain chunks are ranked first |
| Make-up examination arrangement | Make-up examination time, location, and invigilator signature must correspond to the approval result of the deferred examination application and the course record of the transcript | Adjust similarity and limit, record the number of candidates and scores, and check whether the quoted text retains the original time and location |
| Deployment status and operation records | Version, connection, and task records are consistent; if a signature or receipt is missing, the handover is deemed incomplete | Check item by item against the handover form, record the number of candidates and scores, and confirm that the number of quotes and the tokens of quoted text are registered separately |
Three common errors
- Symptom: The name, quantity, and date of the same course do not match between the transcript and the make-up examination arrangement. Cause: The two materials were entered separately, without cross-checking and version synchronization.
- Symptom: The deferred examination application lacks an approval signature or receipt, so the status cannot be confirmed at handover. Cause: The paper archive was not supplemented, and retrieval candidates cover only the body text but not the signature page.
- Symptom: There are many candidates, but the ones ranked first are all irrelevant, and the answer lacks key paragraphs. Cause: The retrieval mode was not selected according to the material form, and re-ranking was not enabled for secondary sorting.
How to verify the setup
This feature's acceptance: test once each with synonym-rewritten questions, compare candidates and scores, then turn re-ranking on and off to observe the order of the top few; at the same time switch searchMode and adjust similarity and limit, recording the number of candidates and scores.
Business test: check item by item against the handover form once to see which record is missing; record the number of quotes separately by array length, and check the tokens of quoted text separately against the budget; do not compare the two for equality.
- In the knowledge base search test, use the three modes semantic, full-text, and hybrid to retrieve the same transcript question, record the number of candidates and scores, and confirm that after multi-path merging both semantic and full-text results are included.
- Run the same question once before and once after enabling re-ranking, compare the order of the top few candidates; then adjust
similarityandlimit, and check whether the quoted text retains the key chunks for course, date, and signature.