What this category's data looks like
This category's data covers three types of materials: achievement registration forms, patent disclosure documents, and software copyright materials. Sources are researcher submissions, agency receipts, and management department ledgers. The update cadence follows application milestones and correction notices. Field characteristics include achievement name, contributor ordering, quantity, date, signature, and attachment receipts. Achievement registration forms carry achievement identity and ownership declarations; patent disclosure documents carry technical solutions and inventor information; software copyright materials carry code versions and rights attribution. When the three types of materials enter parsing and training tasks, each is stored as an independent file record; after parsing and chunking, visible data rows are generated and bound to a unique data identifier, providing a data basis for retrieval and subsequent processing to compare whether the name, quantity, and date of the same object are consistent across two materials, and to supply the discrepancy list.
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. The index transitions through three states: parsed to indexing to indexed. Data pending indexing does not participate in recall. When switching models, only data that has completed indexing is rebuilt.
| Object | Check | Verification action |
|---|---|---|
| Achievement registration form | Whether the name, quantity, and date of the same record are consistent across the two materials, and whether signatures and receipts are complete | Verify the file record and the data row separately, and confirm that chunking under chunkSize and chunkSplitter covers the name, quantity, and date fields |
| Patent disclosure document | Whether inventor ordering and application date match the achievement registration form, and whether versions are in sync | Open the data row to check whether the status is parsed, indexing, or indexed, and compare the disclosure document with the ledger fields |
| Software copyright materials | Whether the software name, version number, and rights attribution are consistent with the registration form; missing signatures or receipts must be flagged | Check the training mode corresponding to trainingType, and confirm that rebuilding only covers data rows that have completed indexing |
| Data row status | Is in one of the three states 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 achievement do not match between the registration form and the disclosure document. Cause: The two materials are filled in by different roles, and after parsing and chunking the fields are not aligned to the same object.
- Symptom: Material versions are not in sync, and an old disclosure document is still hit by retrieval. Cause: Model-switch rebuilding only targets data that has completed indexing; data pending indexing does not participate in the rebuild chain.
- Symptom: Materials missing signatures or receipts still enter the ledger. Cause: Successful parsing does not equal a guaranteed retrieval hit; when the index status has not transitioned to completed, the business side cannot confirm.
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 parsed, indexing, and indexed three states are consistent with the interface display, and that rebuilding only covers data rows that have completed indexing.
Business test: Use one real record to find fields inconsistent with the ledger, verify the file record and the data row separately, locate the name, quantity, or date discrepancy, and hand it to the reviewer for confirmation and write-back.
- Open the data row to check the status, confirm it is in one of parsed, indexing, or indexed, and check whether chunking under
chunkSizeandchunkSplittercovers the name, quantity, and date fields. - View the data rows covered by the rebuild task, confirm it targets only data that has completed indexing, that data pending indexing does not participate in rebuilding, and record the processing mode corresponding to
trainingType.