What this category's data looks like
This category's data looks like this Self-taught examination registration records and graduation review involve three types of materials: registration archives, course qualification certificates, and graduation application forms. Registration archives are filed by the provincial examination office according to examination period, with fields including name, admission ticket number, course code, score, and handler signature; course qualification certificates are issued one by one per course, with fields including course name, credits, qualification date, and seal of the examining institution; graduation application forms are filled in by candidates and signed off level by level, with fields including major name, number of courses, graduation conclusion, reviewer signature, and receipt. The three types of materials update at different rhythms: registration archives roll with the examination period, qualification certificates are reissued by batch, and application forms are submitted intensively during graduation season. When materials enter the document parsing and chunking function, each is stored as a data row as an independent file, and after parsing and chunking is bound to a unique data identifier, so that retrieval and subsequent processing can compare fields, dates, and signatures against the same object.
Configuration and implementation checks
Parsing and training tasks (server-side tasks) only output data rows and index status (parsed, indexing, indexed); the discrepancy list must be handled next by retrieval and subsequent processing. After parsing is complete, 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, and changing the model only rebuilds data that has already been indexed.
| Object | Check | Verification action |
|---|---|---|
| Registration archive | Name, admission ticket number, course code, score, and handler signature are consistent across the two materials | Check the file record and the data row separately, see whether the data row status is parsed, indexing, or indexed, and check whether the chunk boundaries corresponding to chunkSize and chunkSplitter are cut between fields |
| Course qualification certificate | Course name, credits, qualification date, and seal are complete and match the same items in the registration archive | Locate the data rows of the two materials by the same course code, compare the date and quantity fields, and confirm that the training mode corresponding to trainingType has not mixed qualification certificates and application forms into the same index batch |
| Graduation application form | Major name, number of courses, graduation conclusion, reviewer signature, and receipt have no missing items | Check whether the application form data row participates in recall only after it is indexed, check whether the signature and receipt fields are truncated after chunking, and write the discrepancy back to the ledger |
| File and data row | The identities of the file record and the data row are kept separate, and the rebuild scope covers only data that has already been indexed | Check the file record and the data row separately, view the list of data rows covered by the rebuild task, and confirm that data pending indexing has not been included in the rebuild chain |
Three common errors
- Symptom: the number of courses for the same record does not match between the registration archive and the graduation application form. Cause: the versions of the two materials are not synchronized, the old version of the application form was not replaced when the registration record was updated, and after parsing each was stored as an independent data row.
- Symptom: the qualification date on the course qualification certificate is inconsistent with the score date in the registration archive. Cause: the reissued qualification certificate and the original examination period archive have different times, and the two materials were not aligned by the same course code before comparing fields.
- Symptom: the graduation application form is missing the reviewer signature or receipt, and it cannot be confirmed during review. Cause: the signature page of the scanned copy is missing or chunking cut the signature area into an adjacent chunk, and retrieval recall did not cover that field.
How to verify the setup
Acceptance for this function: upload one material, observe the data row status transitions and the coverage scope of the rebuild task, confirm that the status passes through parsed, indexing, and indexed in sequence, and that changing the model rebuilds only data rows that have already been indexed.
Business test: use one real record to find a field that is inconsistent with the ledger, check the file record and the data row separately, write items with mismatched names, quantities, and dates into the discrepancy list, and hand it to the reviewer to confirm and write back.
- Open the data row to check the status, confirm that it is in one of the three states and consistent with the interface display, then compare the name, quantity, and date of the same object in the two materials.
- View the data rows covered by the rebuild task, confirm that only data that has already been indexed is covered, that data pending indexing does not participate in recall, and hand the discrepancy list to the reviewer for confirmation before writing back.