What this category's data looks like
Vocational training subsidy application materials include training class filing records, attendance records, and completion assessment forms. They are sourced from electronic documents and scanned copies submitted by training institutions in batches, and the update cadence rolls with class start and completion milestones. When materials enter the document parsing and chunking feature, the filing form serves as the field definition and terminology baseline, attendance records provide personnel and date details, and completion assessment forms supplement score and signature information. Multilingual materials and terminology handling require that the same term be written consistently throughout and that translated names can refer back to the source text. Therefore, chunking must preserve field boundaries and signature fields so that subsequent retrieval can trace back to the source text by terminology, date, and quantity.
Configuration and implementation checks
Parsing and training tasks (server-side tasks) only output data rows and index status (parsed, indexing, indexed). The unified terminology glossary and comparison draft must be handled next by retrieval and subsequent processing. After parsing is complete, parsed chunks are immediately persisted as visible data rows and bound to a unique data identifier. The index transitions through three states, and data pending indexing does not participate in recall.
| Object | Check | Verification action |
|---|---|---|
| Training class filing record | Whether the name, quantity, date, and signature fields are consistent between the filing form and attendance records | Verify the file records and data rows separately, confirm field boundaries and signature receipts, and check whether terminology remains complete after chunking by chunkSize and chunkSplitter |
| Attendance record | Whether the name, date, and quantity of the same person match between the attendance record and the completion assessment form | Compare the dates and quantities in the two materials one by one, check for missing signatures or receipt items, and view whether the data row status is parsed or indexing |
| Completion assessment form | Whether scores, dates, signatures, and the filing form version are synchronized | Check the version number and signature fields, confirm that only data with completed indexing participates in rebuilding, and check the processing mode corresponding to trainingType |
| Data rows and index status | Whether it is in one of the three states and consistent with the interface display | Open the data row to check the status, confirm that data pending indexing is not recalled, and that the rebuild task covers only indexed data |
Three common errors
- Symptom: The name, quantity, and date of the same record do not match between the filing form and attendance records. Cause: Multilingual material terminology is not unified, field boundaries are lost after chunking, and referring back to the source text fails.
- Symptom: The completion assessment form lacks a signature or receipt, and the version is not synchronized. Cause: Materials are updated in rolling batches, and old-version data rows are still in the completed indexing state.
- Symptom: Newly uploaded materials cannot be found. Cause: Data rows are in parsed or indexing state, and data pending indexing is naturally isolated at the retrieval layer and is not recalled.
How to verify the setup
Acceptance for this feature: 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 rebuilding covers only indexed data.
Business test: Sample three terms to check whether their spelling is consistent throughout, verify that translated names can refer back to the source text, and record the tokens and array count of the cited text.
- Verify the file records and data rows separately, confirm that the unique data identifier is bound correctly, and check the values of
chunkSize,chunkSplitter, andtrainingTypein the chunking and training configuration. - View the data rows covered by the rebuild task, confirm that rebuilding is triggered only for data with completed indexing, and that data pending indexing does not participate in the rebuild chain.