What this category's data looks like
In the scenario of signing scientific research project task books, three types of materials are involved: project task books, subject task breakdown tables, and budget books. The task book provides the research objectives, period, and signature fields; the breakdown table lists subject names, responsible persons, and quantity division; the budget book states subject amounts, dates, and approval receipts. The materials come from the scientific research management archive of this park and are updated with the signing progress; the currently effective version shall prevail. After the three types of materials enter the parsing and training tasks, they each form data rows. The task book and breakdown table are used to compare subject names and dates, and the budget book is used to verify quantities and signatures. Field differences among materials are retained at the data row level for subsequent retrieval to refer back to the original text.
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 basis entries that can refer back to the original text are then completed by retrieval and subsequent processing. After parsing is completed, parsing and chunking are immediately persisted as visible data rows and bound to a unique data identifier. The index transitions through three states, and data pending indexing does not participate in recall.
| Object | Check | Verification action |
|---|---|---|
| Project task book | Fields include subject name, start and end dates, and signature fields, and must match the breakdown table item by item | Check whether the data row status is parsed, indexing, or indexed, and refer back to the signature position in the original text |
| Subject task breakdown table | The subject name, responsible person, and quantity must be consistent with the same record in the task book | Check the file record and the data row separately, and confirm that after chunkSize splitting the entries can match the original text |
| Budget book | Subject amounts, dates, and approval receipts must match the task book period | View the chunkSplitter splitting results, and check entries missing signatures or receipts |
| Data row | It is in one of the three states and consistent with the interface display; rebuilding only covers data whose indexing has been completed | Open the data row to check the status, and confirm trainingType and the coverage scope of the rebuild task |
Three common errors
- Symptom: The name, quantity, and date of the same subject do not match between the task book and the breakdown table. Cause: The two materials are parsed and persisted separately, fields are not compared across materials, and differences remain in their respective data rows.
- Symptom: The budget book version is not synchronized, and retrieval still returns old subject amounts. Cause: The old version's data rows have completed indexing and were not updated in sync with new materials; data pending indexing does not participate in recall.
- Symptom: The task book lacks a signature or receipt, and the basis entry cannot refer back to the complete original text. Cause: Parsing only persists existing content; missing fields do not generate a difference list, and the retrieval layer cannot complete them.
How to verify the setup
Acceptance for this feature: upload one material and observe the data row status transitions and the coverage scope of the rebuild task. Confirm that after parsing is completed it is immediately persisted as visible data rows, with statuses in order parsed, indexing, indexed; data pending indexing is not recalled; changing the model only rebuilds data whose indexing has been completed.
Business test: ask a cross-material basis question, see whether the answer gives a referable source, and check that the cited text's tokens and array item count are recorded separately.
- Open the data row to check whether the status is consistent with the interface display, and confirm that the identities of the file record and the data row are distinguishable.
- View the data rows covered by the rebuild task, and confirm that it only covers data whose indexing has been completed; the initial training covers data that is pending indexing and indexing.