What this category's data looks like
This category's data for compulsory education dropout prevention and control includes three types of materials: dropout risk ledgers, return-to-school persuasion records, and home visit records. They are submitted by schools and townships and updated by semester and by month. Ledger fields include student name, risk level, date, and signature; persuasion records include number of persuasion attempts, receipt, and signature; home visit records include home visit date, teacher signature, and parent signature. When the materials enter the document parsing and chunking function, the ledger serves as the main index object, while persuasion records and home visit records serve as supporting materials. Each is bound to a unique data identifier and persisted as a visible data row, for retrieval and subsequent processing to generate evidence entries that can be traced back to the original text, used to answer parents or write into plans.
Configuration and implementation checks
This step is executed by the parsing and training task (server-side task), outputting data rows and index status (parsed, indexing, indexed). Evidence entries that can be traced back to the original text are then completed by retrieval and subsequent processing. After parsing is completed, parsing and chunking are immediately persisted 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 been completed.
| Object | Check | Verification action |
|---|---|---|
| Dropout risk ledger | Fields, dates, and signatures are complete, and the names, quantities, and dates match those in the persuasion records and home visit records | Check the student name, risk level, date, and signature in the ledger, and compare them item by item against the corresponding entries in the persuasion records and home visit records |
| Persuasion record | Receipt and signature are complete, the version is the current one, and entries can be traced back to the original text | Check the number of persuasion attempts, receipt, signature, and date, and check whether the name, quantity, and date of the same record are consistent across the two materials |
| Home visit record | Teacher and parent signatures are complete, the date is consistent with the ledger, and the version has been synchronized | Check the home visit date, teacher signature, and parent signature, check for missing signatures or receipts, and confirm that the rebuild scope covers only data whose indexing has been completed |
| Data row status | It is in one of the three states and consistent with the interface display; chunkSize, chunkSplitter, and trainingType are consistent with the task record | Open the data row and check the status parsed / indexing / indexed, and check the respective identity identifiers of the file record and the data row |
Three common errors
- Symptom: materials that were just uploaded cannot be found. Cause: the data row is in parsed or indexing state; data pending indexing does not participate in recall, and successful parsing does not mean retrieval will necessarily hit.
- Symptom: the name, quantity, and date of the same record do not match across the two materials. Cause: the ledger and the persuasion records and home visit records have not been version-synchronized, and the field definitions are inconsistent.
- Symptom: an empty data row is left after training fails. Cause: materials missing signatures or receipts were mistakenly entered into the training chain; the initial training covers data pending indexing and data being indexed, while rebuild only targets data whose indexing has been completed.
How to verify the setup
This function's acceptance: upload one material and observe the data row status transitions and the coverage scope of the rebuild task. Check whether the three states parsed, indexing, and indexed are consistent with the interface, check whether chunkSize, chunkSplitter, and trainingType are consistent with the task record, and confirm that rebuild covers only data whose indexing has been completed.
Business test: ask a cross-material evidence question and see whether the answer provides a traceable source. Check that the evidence comes from this institution's materials, that the version is the current one, and that entries match the original text, and check that the tokens and array count of the quoted text are recorded separately.
- Open the data row and check the status parsed / indexing / indexed, and separately check the respective identity identifiers of the file record and the data row.
- Check the data rows covered by the rebuild task, confirm that only data whose indexing has been completed is covered, and that data pending indexing does not participate in the rebuild chain.