What this category's data looks like
This page focuses on business basis queries in daily life management in kindergarten childcare. The materials come from the kindergarten's current daily life schedule, morning and afternoon check records, child health observation records, and the accompanying disinfection and handover receipts. The three types of materials are updated daily or weekly, and their fields include class, child name, time node, activity name, observation indicator, handling opinion, date, and signature field. When entering the document parsing and chunking function, the daily life schedule provides schedule and segment fields, the morning and afternoon check records provide health screening and abnormality markers, the child health observation records provide continuous observation and follow-up descriptions, and the receipts and signature fields provide clues for responsibility confirmation. Chunking is split by heading level and table row, retaining field names and dates, so that subsequent retrieval can locate the specific material, specific date, and specific signature position, forming basis entries that can be traced back to the original text.
Configuration and implementation checks
This step is executed by the parsing and training task (server-side task), which outputs data rows and index status (parsed, indexing, indexed). "Basis entries that can be traced back to the original text" are then completed by retrieval and subsequent processing. After parsing is completed, parsed chunks are immediately stored as visible data rows and bound to a unique data identifier. The index transitions through three states. Data pending indexing does not participate in recall. Changing the model only rebuilds data whose indexing has already been completed.
| Object | Check | Verification action |
|---|---|---|
| Daily life schedule | The fields include class, segment name, start and end time, organizer, and date. During verification, check whether the segment name and time match the corresponding time period in the morning and afternoon check records, and whether the signature field is complete. | Open the data row and check whether the status is parsed, indexing, or indexed, and check whether chunkSize, chunkSplitter, and trainingType match the material form. |
| Morning and afternoon check records | The fields include child name, check time, temperature or health marker, handling opinion, and signature. During verification, check whether the name and date for the same child on the same date match across the two materials. | Verify the identifiers of the file record and the data row separately, confirm that the identities of the file record and the data row can be distinguished from each other, and then check whether the data rows covered by the rebuild task fall only on data whose indexing has already been completed. |
| Child health observation records | The fields include observation date, observation indicator, description, follow-up measures, and signature receipt. During verification, check whether the description entries match the original paragraphs and whether the receipt is missing. | Open the data row and check whether the status is consistent with the interface display, confirm that data pending indexing is not recalled, and that training-failed data can be directly edited or deleted. |
Three common errors
- Symptom: The name or date for the same child on the same date does not match between the morning and afternoon check records and the health observation records. Cause: After chunking, the fields of the two materials are not aligned, retrieval hits different fragments, and trace-back points to different original text positions.
- Symptom: The daily life schedule has been updated to a new version, but the morning and afternoon check records still reference the old segment name. Cause: The versions are not synchronized, the old data row is still in the completed indexing state, and retrieval preferentially recalls the old fragment.
- Symptom: The health observation records lack a signature or receipt, and the basis entry cannot confirm responsibility. Cause: During chunking, the signature field was split into an adjacent fragment, and when tracing back to the original text, that field is not within the same data row.
How to verify the setup
This function's acceptance: upload one material, observe the data row status transitions and the coverage scope of the rebuild task, confirm that the status only transitions among parsed, indexing, and indexed, that rebuild only covers data rows whose indexing has already been completed, and check chunkSize, chunkSplitter, trainingType, the number of citations, and the citation content budget are recorded separately.
Business test: ask a cross-material basis question and see whether the answer provides a traceable source.
- Open the data row and check whether the status is consistent with the interface display, confirm that data pending indexing does not participate in recall, and that training-failed data can be directly edited or deleted.
- Verify the identifiers of the file record and the data row separately, and check whether the data rows covered by the rebuild task fall only on data whose indexing has already been completed.