What this category's data looks like
In the scenario of investigating and handling academic misconduct reports, materials fall into three categories: report materials, investigation reports, and expert appraisal opinions. Sources include real-name or anonymous report letters, investigation team stage reports, and appraisal documents signed by external experts. The update cadence is driven by four stages: acceptance, verification, panel deliberation, and feedback. Each time evidence is supplemented or conclusions are revised, a new version is produced. Field characteristics center on record name, quantity, date, and signature receipt: the name, quantity, and date of the same record often do not match across two materials; when versions are not synchronized, the old version is still referenced; missing signatures or receipts make conclusions untraceable. The three material categories play different roles when entering this feature: report materials serve as the original source of facts, investigation reports serve as the carrier of process conclusions, and expert appraisal opinions serve as the basis for professional judgment. All must be parsed and chunked, then persisted as data rows for subsequent retrieval and version ledger maintenance.
Configuration and implementation checks
Parsing and training tasks (server-side tasks) only output data rows and index status (parsed, indexing, indexed). The version ledger and change records 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: parsed, indexing, indexed. Data pending indexing does not participate in recall. Changing the model only rebuilds data that has completed indexing.
| Object | Check | Verification action |
|---|---|---|
| Report materials | Whether fields, dates, and signatures are consistent with the investigation report, and whether receipts are complete | Open the data row to check the status, compare name, quantity, date, and signature items, and record the version number |
| Investigation report | Whether the version is unique, whether the old version is traceable, and whether references are synchronized | Check the data rows covered by the rebuild task, confirm that only indexed data is covered, and verify the recorded trainingType and chunkSize settings |
| Expert appraisal opinions | Whether signatures and receipts are complete, and whether the appraisal opinions correspond to the report conclusions | Verify the file record and the data row separately, confirm that the file identifier and the data row identifier are distinguishable, and inspect the chunkSplitter chunking results |
| Data row | Is in one of the three states parsed / indexing / indexed and consistent with the interface display | Open the data row to check the status, and confirm that data pending indexing does not participate in recall |
Three common errors
- Symptom: The name, quantity, and date of the same record do not match between the report materials and the investigation report. Cause: The two materials are parsed and persisted separately, and versions are not aligned by unique data identifier.
- Symptom: After the investigation report is updated, the old version is still hit by retrieval, and references are not synchronized. Cause: The version ledger is not updated as the index status transitions, and the old version's data row is still in a recallable state.
- Symptom: Expert appraisal opinions lack signatures or receipts, and conclusions cannot be traced. Cause: Parsing and chunking only persist the body text, and signature and receipt fields are not included in data row verification.
How to verify the setup
Feature acceptance: Upload one material, observe the data row status transitions and the coverage scope of the rebuild task, confirm that the status correctly transitions among parsed, indexing, and indexed, that rebuild only covers indexed data, and verify the records of the three identifiers trainingType, chunkSize, and chunkSplitter.
Business test: Change one piece of content and see whether the version number and references are updated synchronously.
- Open the data row to check the status, confirm it is consistent with the interface display, and verify the file record and the data row identifier separately.
- Check the data rows covered by the rebuild task, confirm that only data that has completed indexing is covered, and that data pending indexing does not participate in rebuild.