What this category's data looks like
This page addresses the scenario of pre-departure training and overseas safety reminders. Materials include pre-departure training courseware, safety reminder checklists, and accommodation contact sheets. Sources are internal institutional teaching and research plus overseas service records. The update cadence adjusts with semester batches, destination policies, and contact person changes. Field characteristics center on names, dates, quantities, signatures, and acknowledgments. The same record often appears simultaneously in the courseware and the checklist. When materials enter knowledge base retrieval and recall, the courseware handles process explanations and case illustrations, the safety reminder checklist handles itemized reminders, and the accommodation contact sheet handles contact and emergency information. The version numbers, effective dates, and reference relationships of the three must remain consistent. Old versions are retained for traceability, the current version is unique, and references are updated in sync, avoiding mismatches in names, quantities, and dates for the same record across the two materials.
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 pattern, merge, re-rank, filter, and upper-limit trimming). The version ledger and change records are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Pre-departure training courseware | Whether fields, dates, and signatures are consistent with the safety reminder checklist | In the search test, test once each with synonymous questions, record the number of candidates and scores, and check the names, dates, quantities, and signature fields in the courseware |
| Safety reminder checklist | Whether item names, quantities, and effective dates match the accommodation contact sheet | Switch searchMode to semantic, full-text, or hybrid, compare whether candidates simultaneously contain snippets from both types of materials, and check the checklist items against the contact sheet records |
| Accommodation contact sheet | Whether contact person, dates, and acknowledgments are missing, and whether old versions are traceable | Compare the ranking of the top few items before and after enabling usingReRank, filter by similarity and then check the tokens of the referenced text, record the array count separately |
| Version ledger and change records | Current version is unique, old versions are traceable, references are in sync | After changing one piece of content, re-run retrieval to see whether the version number and references are updated in sync, and check whether key snippets are retained after limit trimming |
Three common errors
- Symptom: The name, quantity, and date of the same record do not match between the courseware and the checklist. Cause: The two materials are maintained separately and are not aligned by unified fields and effective dates.
- Symptom: Old versions are still recalled, and the current version is not unique. Cause: Versions are not synchronized, the ledger does not record changes, and references are not updated with content.
- Symptom: The accommodation contact sheet is missing signatures or acknowledgments. Cause: The material circulation step did not collect all acknowledgments, and fields were left blank before directly entering the knowledge base.
How to verify the setup
This feature acceptance: test once each with synonym-rewritten questions, compare candidates and scores, then toggle re-ranking to see the ranking of the top few items, and check the effects of similarity, searchMode, usingReRank, and limit.
Business test: change one piece of content and see whether the version number and references are updated in sync.
- In the knowledge base search test, switch
searchModeand compare the number of candidates and scores across semantic, full-text, and hybrid modes. - Compare the order of the top few items before and after enabling
usingReRank, filter bysimilarityand then check the tokens of the referenced text, record the array count separately, and check whether key snippets are retained afterlimittrimming.