What this category's data looks like
This page is intended for enterprise compliance and security training scenarios. Materials include three categories: compliance training courseware, security operating procedures, and exam records. Courseware and procedures are usually issued by internal policy departments on an annual basis or according to the pace of regulatory updates, while exam records are generated by the training organizer in batches, with fields covering course name, clause number, training date, trainee name, quantity, signature, and receipt status. When materials enter this feature, courseware and procedures serve as baseline materials for parsing and chunking and are stored as data rows, while exam records serve as comparison materials and are likewise parsed and chunked; the two are retrieved and have their fields aligned during retrieval and subsequent processing stages. In terms of update cadence, regulatory changes trigger revisions to courseware and procedures, exam records are added with each training session, and version differences and missing signatures are common verification priorities.
Configuration and implementation checks
This step is executed by parsing and training tasks (server-side tasks), outputting data rows and index status (parsed, indexing, indexed). The "discrepancy list" is then completed by retrieval and subsequent processing.
| Object | Check | Verification action |
|---|---|---|
| Compliance training courseware | The course name, clause number, and training date in the courseware match the corresponding fields in the exam records | Open the data row verification status, confirm the trainingType and chunkSplitter settings, and compare the course name and date fields in the courseware and exam records |
| Security operating procedures | The procedure version number, effective date, and signature column are synchronized with the versions referenced by the courseware and exam records | Verify the file records and data rows separately, check the impact of chunkSize on chunk boundaries, and confirm the version number and signature receipt |
| Exam records | The trainee name, quantity, exam date, signature, and receipt status are consistent with the courseware and procedures | Retrieve the exam record data rows, compare the name, quantity, and date fields, and mark items missing signatures or missing receipts |
| Data row status | It is in one of the three states and is consistent with the interface display | Open the data row verification status, and confirm that the parsed, indexing, and indexed transitions are consistent with the interface display |
Three common errors
- Symptom: The same trainee's name or quantity does not match between the courseware and the exam records. Cause: The two materials are entered separately without field-level alignment, and after parsing and chunking they are stored independently.
- Symptom: The procedure has been revised, but the exam records still reference clauses from the old version. Cause: The versions are not synchronized, and after the courseware and procedures are updated, re-parsing of the comparison materials is not triggered.
- Symptom: Exam records lack a signature or receipt but are still treated as valid records. Cause: The signature and receipt fields are not included in the verification dimensions, and missing items are not marked after parsing and storage.
How to verify the setup
Feature acceptance: Upload one material, observe the data row status transitions and the coverage scope of the rebuild task, and confirm that parsing and chunking are immediately stored as visible data rows after parsing is completed and bound to a unique data identifier; the index transitions through the three states parsed, indexing, and indexed, and changing the model only rebuilds data whose indexing has already been completed.
Business test: Use one real record to find fields inconsistent with the ledger, and compare the name, quantity, date, and signature receipt in the courseware, procedures, and exam records.
- Open the data row verification status, confirm that it is in one of parsed, indexing, or indexed and consistent with the interface display, and that data pending indexing does not participate in recall.
- Verify the file records and data rows separately, confirm that the identities of the file records and data rows are distinguishable from each other, and check whether the data rows covered by the rebuild task include only data whose indexing has already been completed.