What this category's data looks like
This page addresses the technical college formal education management scenario, and the materials come from three categories: student registration records, teaching plans, and skills training records. Student registration records are updated at semester registration nodes, and their fields include student name, student ID, major, class, registration date, and handler signature; teaching plans are revised on a rolling basis by major and grade, and their fields include course name, number of class hours, semester offered, and approval signature; skills training records are generated by week or work order, and their fields include training item, equipment number, completed quantity, date, and supervisor signature. When the materials enter the knowledge base retrieval and recall function, student registration records play the role of verifying identity and enrollment status, teaching plans play the role of verifying course and class-hour definitions, and skills training records play the role of verifying the training process and results. Together, the three support exception tracing and manual review; retrieval results are used to locate original record fragments, and review conclusions are signed and confirmed manually after comparing candidate fragments with the original materials.
Configuration and implementation checks
This step is performed by the knowledge base retrieval service (retrieval node and search test), which outputs candidate fragments and similarity scores (after mode, merge, rerank, filter, and limit trimming). The "exception list and review conclusions" are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Student registration records | Whether name, student ID, major, class, registration date, and signature are consistent between the registration records and the teaching plans | In the knowledge base search test, switch searchMode to semantic, full-text, or hybrid, enter the student name and student ID, check the registration fields and signature in the returned candidate fragments, and record the number of candidates and the similarity score |
| Teaching plans | Whether course name, number of class hours, semester offered, and approval signature match between the plans and the skills training records | Search using a combination of course name and semester, compare the top few rankings before and after enabling usingReRank, check the class-hour count and course start date fields, and record the number of candidates and the score |
| Skills training records | Whether training item, equipment number, completed quantity, date, and supervisor signature are consistent between the training records and the teaching plans | Enter the training item and equipment number to search, check the completed quantity and date in the returned fragments, check whether the signature or receipt is missing, record the number of candidates and the similarity score, and if necessary adjust limit to observe changes in the number of cited items |
Three common errors
- Symptom: the name or student ID of the same student is written inconsistently in the registration records and the training records, so tracing locates different records. Cause: the two entries were not checked against the same field definitions, and names and quantities were not cross-compared.
- Symptom: after the teaching plan is revised, the number of class hours or the semester offered changes, but the training records are still executed according to the old version. Cause: the versions were not synchronized, and retrieval hit old-version fragments without checking dates and signatures against the new plan.
- Symptom: the training records lack a supervisor signature or receipt, so the review conclusion cannot be signed and confirmed. Cause: the original record collection step was not closed-loop, the signature field in the returned retrieval fragments is empty, and manual review lacks a basis.
How to verify the setup
Feature acceptance: test once each with synonym-rewritten questions, compare candidate fragments and similarity scores, then toggle usingReRank on and off to see changes in the ranking of the top few items, and check whether the candidate sources after switching searchMode include both semantic and full-text results.
Business test: create one exception record and see whether the source material can be traced.
- In the knowledge base search test, fix on one student or one training record, search separately using the three
searchModeoptions semantic, full-text, and hybrid, record the number of candidates and scores, and check whether the returned fragments cover the corresponding fields and signatures of the student registration records, teaching plans, and skills training records. - For the same exception record, search once with
usingReRankenabled and once disabled, and compare the order of the top few candidates; then adjustlimitto observe changes in the number of cited items, separately record the tokens of the cited text and the number of array items, and check whether dates, quantities, and names are consistent with the original materials.