What this category's data looks like
Vocational school student internship management materials usually include three types: internship agreements, internship plans, and internship logs. The agreement is signed by three parties—the school, the enterprise, and the student—and contains fields such as position, term, remuneration, and insurance, as well as signature fields; the plan is prepared by the professional teaching and research office and contains internship objectives, rotation arrangements, supervising teachers, and date ranges; the log is filled in by the student on a daily basis and contains tasks, work hours, the signature of the mentoring master, and a weekly receipt. When the three types of materials enter knowledge base retrieval, their roles differ: the agreement and plan serve as sources of basis entries, while the log serves as process evidence. Materials are updated by semester or internship batch, and versions are distinguished by signature date and document number. Common field characteristics: for the same record, the position name, number of people, and start and end dates in the agreement and plan must be consistent; if the log lacks a signature or receipt, it cannot serve as a valid basis. Retrieval must be able to refer back to the specific material and clause, for replying to parents or writing into a plan.
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). Basis entries that can refer back to the original text are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Internship agreement | Whether the position, term, remuneration, and insurance fields and signature fields in the agreement correspond to the internship plan, and whether the dates and number of people match | In the search test, switch searchMode to hybrid, ask whether the agreement position and the plan position are consistent, and see whether the candidates simultaneously provide snippets and scores from both materials |
| Internship plan | Whether the rotation arrangements, supervising teachers, and date ranges in the plan are consistent with the log records, and whether the version is the current one | Test once each with synonymous question sentences, compare the number of candidates and scores, enable usingReRank and see whether the top few are ranked higher |
| Internship log | Whether the log's tasks, work hours, mentoring master's signature, and weekly receipt are complete, and whether entries missing a signature or receipt can be identified | Adjust the similarity threshold to observe whether low-scoring candidates are filtered out, record the tokens of the quoted text, and record the number of array entries separately |
| Knowledge base retrieval service | Whether candidate snippets and similarity scores are output after pattern, merge, re-rank, filter, and upper-limit trimming, and whether the quoted content retains key snippets of the original text | Check whether the returned candidates simultaneously contain semantic and full-text results, and verify the number of quoted entries and content completeness after limit trimming |
Three common errors
- Phenomenon: the same position name is written differently in the agreement and the plan, and retrieval hits only one. Cause: material naming is not unified, and hits and associations were not checked first.
- Phenomenon: the plan has changed to a new version, but the log is still filled in according to the old version, and the dates do not match. Cause: versions are not synchronized, and retrieval does not filter by the current version.
- Phenomenon: the log lacks the mentoring master's signature or weekly receipt but is still cited as a basis. Cause: field completeness is not validated, and similarity filtering does not cover it.
How to verify the setup
This feature's acceptance: test once each with synonymously rewritten question sentences, compare candidates and scores, then toggle re-rank on and off to see the ranking of the top few; record the number of candidates and scores, and do not write the three states of indexing.
Business test: ask a basis question that spans materials and see whether the answer gives a traceable source.
- In the search test, switch
searchModeto semantic, full-text, and hybrid, compare the number of candidates and scores, and confirm the multi-path merged results. - Enable
usingReRankto compare the ranking of the top few, adjustsimilarityandlimit, and verify the tokens of the quoted text and the number of array entries.