What this category's data looks like
This category's data involves three core types of materials: registration forms, care bonus-point materials, and admission rosters. Registration forms are filled in by candidates and contain name, ID number, application level, subject category, signature, and date; care bonus-point materials are issued by candidates or relevant departments and contain certificate name, quantity, issue date, handler signature, and receipt; admission rosters are generated by admissions institutions and contain candidate name, admission quantity, admission date, and responsible person's signature. Materials are submitted centrally by registration season and updated in batches; fields are mainly name, quantity, date, and signature, and the same record must be compared item by item across the three materials. After materials enter the document parsing and chunking function, each becomes an independent data row; registration forms are used to verify candidates' basic fields, care bonus-point materials are used to verify the basis for bonus points and receipts, and admission rosters are used to verify admission results and signatures. The three support discrepancy localization under the same index status.
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). The discrepancy list is then completed by retrieval and subsequent processing. After parsing is complete, parsing and chunking are immediately persisted as visible data rows and bound to a unique data identifier; the index transitions through the three states parsed, indexing, and indexed, and data pending indexing does not participate in recall.
| Object | Check | Verification action |
|---|---|---|
| Registration form | Name, quantity, date, and signature must be consistent item by item with the care bonus-point materials and admission roster | Verify the file record and data row separately, confirm the data row status is parsed, indexing, or indexed, and check whether the chunk boundaries corresponding to chunkSize and chunkSplitter cover the signature and date fields |
| Care bonus-point materials | Certificate name, quantity, issue date, and handler signature must be consistent with the registration form | Open the data row to verify the status, compare the receipt and signature fields, and confirm that the training mode corresponding to trainingType matches the material type |
| Admission roster | Candidate name, admission quantity, admission date, and responsible person's signature must be consistent with the registration form and bonus-point materials | Verify the file record and data row identifier, and check whether the data rows covered by the rebuild task contain only data with completed indexing |
| Data row status | Is in one of the three states parsed, indexing, or indexed and is consistent with the interface display | Open the data row to verify the status, confirm that data pending indexing is not recalled, and that data with failed training can be edited or deleted |
Three common errors
- Symptom: The same candidate's name, quantity, and date do not match between the registration form and the admission roster. Cause: The two materials were parsed and persisted separately, fields were not compared item by item, and the discrepancy was not localized to a specific data row.
- Symptom: The version of the care bonus-point materials is not synchronized, and the old receipt and the new certificate coexist. Cause: When materials were updated in batches, they were not merged by unique data identifier, and the index status did not reflect the latest version.
- Symptom: The admission roster lacks the responsible person's signature or the registration form lacks the candidate's receipt. Cause: The signature and receipt fields were not included in the chunk verification scope, and missing items were not marked after parsing.
How to verify the setup
This feature's acceptance: Upload one material, observe the data row status transitions and the coverage scope of the rebuild task, confirm that the transitions from parsed to indexing to indexed are consistent with the interface display, and that the rebuild covers only data with completed indexing.
Business test: Use one real record to find fields inconsistent with the ledger, verify the file record and data row separately, locate discrepancies in name, quantity, date, or signature, submit them to the reviewer for confirmation, and write them back.
- Open the data row to verify the status, confirm that the three-state transitions are consistent with the interface, and record the chunking and training information corresponding to
chunkSize,chunkSplitter, andtrainingType. - Verify the file record and data row separately, check whether the data rows covered by the rebuild task contain only data with completed indexing, and submit the discrepant fields to the reviewer for confirmation and write-back.