What this category's data looks like
In the grade entry and credit recognition scenario, three types of materials—transcripts, deferred examination applications, and make-up examination arrangements—enter knowledge base retrieval at the same time. Transcripts carry course names, credits, grades, and entry dates; deferred examination applications carry the original examination date, the reason for deferral, and approval signatures; make-up examination arrangements carry the make-up examination time, location, invigilation, and receipt. The three types of materials update at different rhythms: transcripts roll forward with entry batches, deferred examination applications are filled in as approvals circulate, and make-up examination arrangements are adjusted with academic scheduling. When entering this feature, the three serve respectively as candidate sources for semantic recall, full-text recall, and hybrid recall, used to compare whether the name, quantity, date, and signature of the same record are consistent, and to support field verification before material drafting and responsibility handover.
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 pattern, merge, re-rank, filter, and limit trimming). The material drafts and handover records that can be issued are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Transcript | Fields complete, responsible person clear, version and date consistent | In the search test, switch searchMode to semantic, full-text, or hybrid, check the number of candidates and scores, and confirm that course name, credits, grades, and entry date are complete |
| Deferred examination application | Fields complete, responsible person clear, version and date consistent | Compare the ordering of the first few items before and after enabling usingReRank, and check whether the original examination date, reason for deferral, and approval signature are consistent with the transcript |
| Make-up examination arrangement | Fields complete, responsible person clear, version and date consistent | Adjust similarity to observe changes in the number of candidates and scores, check the make-up examination time, location, invigilation, and receipt, and confirm that version and date are synchronized |
| Handover record | Responsible person clear, version and date consistent | Check whether the candidate chunks cover both parties to the handover, the handover date, and the receipt; when items are missing, go back to hits and associations for troubleshooting |
Three common errors
- Symptom: the name and credits of the same course do not match between the transcript and the deferred examination application. Cause: the two materials were drafted separately, without verifying fields and dates against the same record.
- Symptom: the make-up examination arrangement has been updated, but the handover record still references the old version. Cause: the version was not synchronized, the retrieval candidates did not cover the latest chunks, and dates were not compared before handover.
- Symptom: the deferred examination application lacks an approval signature or receipt, but the material still proceeds to the next stage. Cause: field verification did not block the missing items, and the responsible person and receipt status were not stated in the draft.
How to verify the setup
Feature acceptance: test once each with synonym-rewritten questions, compare the number of candidates and scores, then toggle re-ranking on and off to inspect the ordering of the first few items; record the number of array items, verify the tokens of the referenced text separately, and do not compare the number of items with the citation budget as equal.
Business test: draft one according to the template and complete one handover, and see whether missing items are blocked.
- In the knowledge base search test, switch retrieval modes, compare the number of candidates and scores, and confirm that the returned candidates include both semantic and full-text results.
- Compare the ordering of the first few items before and after enabling re-ranking, and check whether key chunks of the original text are still retained after similarity filtering and citation limit trimming.