What this category's data looks like
This page addresses the scenario of student status establishment and changes in primary and secondary schools. Materials include three categories: student status records, transfer applications, and leave-of-absence/resumption materials. Student status records contain students' basic information, guardians, enrollment history, and change records, with many fields and both signature fields and date fields; transfer applications contain information about the original school and the receiving school, the reason for transfer, opinions from both parties, and the receipt; leave-of-absence/resumption materials contain the reason for application, duration, approval opinions, and resumption verification. The three categories of materials update at different rhythms: student status records update immediately with semesters and changes, transfer applications are generated per instance, and leave-of-absence/resumption materials are generated by reason cycle. When materials enter knowledge base retrieval, each serves as a source of candidate segments, playing a reference role for field comparison and version tracing, for subsequent drafting and handover verification.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate segments 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 |
|---|---|---|
| Student status record | Fields are complete, responsible persons are clear, version and date are consistent; check whether name, student status number, change date, and signature field are consistent with the transfer application | In the search test, retrieve once each with synonymous questions, record the number of candidates and scores, check whether the returned segments cover the signature and date fields, and switch searchMode to compare the candidate differences among semantic, full-text, and hybrid |
| Transfer application | The names of the original school and receiving school, transfer date, opinions from both parties, and receipt are complete; they match the same record's name and date in the student status record | Turn usingReRank on and off and view the top few rankings each time to confirm whether relevant segments are ranked first; check whether the tokens of the returned reference text are complete, and record the array count separately |
| Leave-of-absence/resumption materials | The reason for application, leave start and end dates, approval opinions, and resumption verification fields are complete; the version and date are consistent with the record | Adjust similarity to observe changes in the number of candidates and scores before and after candidates are filtered, and check whether the returned segments retain the key sentences of the original text; confirm the number of references and content completeness after limit trimming |
Three common errors
- Symptom: The name or date of the same student does not match between the student status record and the transfer application. Cause: The two materials were drafted separately, the same record was not checked back, and the retrieval candidates did not cover the corresponding fields.
- Symptom: The version of the leave-of-absence/resumption materials is not synchronized, and the approval opinions are inconsistent with the latest record. Cause: During drafting, the version and date were not checked against the latest candidate segments, and old segments were directly reused.
- Symptom: The transfer application lacks the original school's signature or receipt and is returned during handover. Cause: The retrieval candidates did not cover the signature field and receipt field, and the drafting step did not check item by item according to the template.
How to verify the setup
This feature's acceptance: retrieve once each with synonymously rewritten questions, compare the number of candidates and scores, then turn re-ranking on and off to view the top few rankings; switch searchMode to check the candidate differences among semantic, full-text, and hybrid, and adjust similarity and limit to view the reference text tokens after filtering and trimming, recording the array count separately.
Business test: draft one according to the template and complete one handover to see whether missing items are blocked.
- In the search test, retrieve once each using different phrasings of the same question, record the number of candidates and scores, and check whether the returned segments cover the signature, date, and version fields.
- Switch
searchModeandusingReRank, compare candidate ranking and content completeness, and confirm that the number of references and reference text tokens are recorded separately.