What this category's data looks like
In the exam question setting and grading category for primary and secondary schools, materials include the question-setting two-way specification table, exam papers, scoring rubrics, and supporting ledgers. Sources are teaching and research groups, question-setting groups, and school administrative archives, and the update cadence rolls with exam batches and revised versions. When these materials enter the document parsing and chunking feature, the specification table plays the role of mapping assessment points to question types, the exam paper plays the role of question statements and score values, and the scoring rubric plays the role of scoring points and signature receipts. Field characteristics center on the name, quantity, date, version number, and signature column of the same object. After parsing, they are split by structure into retrievable data rows for subsequent verification and discrepancy localization.
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.
| Object | Check | Verification action |
|---|---|---|
| Question-setting two-way specification table | The name, quantity, and date of the same assessment point are consistent across the two materials, and the signature column is complete | Verify that the data row status is parsed, then trigger indexing, and confirm that chunkSize chunking has not broken the assessment point name and quantity fields |
| Exam paper | Question numbers, score values, and dates align item by item with the specification table, and version numbers are synchronized | Verify the file record and the data row separately, check the processing mode corresponding to trainingType, and confirm that it does not participate in recall during indexing |
| Scoring rubric | The number of scoring points, dates, and signatures or receipts are consistent with the ledger | Look at the data rows covered by the rebuild task, and confirm that only indexed data is rebuilt; parsed and indexing do not enter the rebuild chain |
| Ledger | The name, quantity, and date of the same record do not contradict the above materials | Use the chunkSplitter result to locate the discrepancy fields, hand them to the reviewer for confirmation, and write back |
Three common errors
- Symptom: The name, quantity, and date of the same record do not match between the specification table and the exam paper. Cause: The two material versions are not synchronized, and after parsing they are stored separately without field comparison.
- Symptom: The scoring rubric lacks a signature or receipt, so review cannot be closed. Cause: The signature column was not validated when the material entered parsing; although the data row status is parsed, business requirements are missing.
- Symptom: Newly uploaded materials cannot be found. Cause: The data row is in parsed or indexing, and data awaiting indexing is naturally isolated at the retrieval layer and is not recalled.
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 the transition from parsed to indexing to indexed, and that rebuild only covers indexed data rows.
Business test: Use one real record to find fields inconsistent with the ledger, form a discrepancy list, hand it to the reviewer for confirmation, and write back.
- Open the data row verification status, confirm it is consistent with the interface display, and verify the identity identifiers of the file record and the data row separately.
- View the data rows covered by the rebuild task, and confirm that it targets only data whose indexing is complete; data awaiting indexing does not participate in the rebuild chain.