What this category's data looks like
This page addresses kindergarten operation behavior and quality self-assessment materials, covering self-assessment reports, issue lists, and rectification plans. Sources are kindergarten submissions, district/county consolidation, and responsible inspector receipts, with updates rolling according to semester evaluation and rectification cycles. When entering knowledge base retrieval, the self-assessment report serves as the main narrative material providing descriptions of indicator achievement, the issue list carries item-by-item issues and responsibility assignments, and the rectification plan records measures, deadlines, and signature status. The three types of materials often contain multilingual parallel terminology, such as operating conditions, safety and hygiene, care and education, faculty and staff, and internal management, requiring unified translations and ensuring traceability back to the original text. Field verification focuses on names, quantities, dates, and signatures, avoiding the same record mismatching across 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, rerank, filter, and limit trimming). The unified terminology glossary and parallel draft are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Self-assessment report | Indicator names, achievement descriptions, dates, and signatures must correspond item by item with the issue list and rectification plan | In the search test, use synonymous questions to retrieve the same indicator, record the number of candidates and scores, and check whether the returned snippets contain the original indicator text |
| Issue list | Issue item names, quantities, responsible persons, and rectification deadlines must be consistent with the rectification plan | Switch searchMode to semantic, full-text, or hybrid, and compare whether candidates cover both issue descriptions and rectification measures |
| Rectification plan | Measures, completion dates, signatures, or receipts must be consistent with the self-assessment report and issue list | Enable and disable usingReRank, compare changes in the top few rankings, and record candidates retained after similarity filtering |
| Multilingual terminology parallel draft | The same term is written consistently throughout, and translations can be traced back to the original text | Sample three terms to check their writing throughout, and verify whether candidate snippets retain key original paragraphs and corresponding terms |
Three common errors
- Phenomenon: The same issue cannot be recalled when rephrased. Cause: Insufficient question optimization, unresolved references, and semantic and full-text candidates not fully merged.
- Phenomenon: The same record's name, quantity, and date do not match between the self-assessment report and the rectification plan. Cause: Versions are not synchronized, and materials are not re-entered into retrieval after updates.
- Phenomenon: The issue list lacks signatures or receipts, but the rectification plan shows completion. Cause: The two materials are maintained separately and not verified against the same fields.
How to verify the setup
This feature acceptance: Test once each with synonymously rewritten questions, compare candidates and scores, then toggle rerank to see the top few rankings; verify candidate counts under searchMode, usingReRank, similarity, and limit, and for citation limits only check citation text tokens, recording array counts separately.
Business test: Sample three terms to check whether their writing is consistent throughout, and verify whether translations can be traced back to the original text.
- In the search test, switch retrieval modes, record candidate counts and scores, and confirm that semantic and full-text results are merged.
- Compare the order of the top few before and after enabling rerank, and verify that key original snippets are still retained after similarity filtering.