What this category's data looks like
Laboratory hazardous chemical procurement and ledgers involve three types of materials: procurement applications, usage registrations, and inventory ledgers. Sources include supplier accompanying documents, internal forms, and historical ledgers, with update cadence advancing along with batch arrivals, usage, and monthly inventory counts. Field characteristics center on product name, specification, quantity, date, and signed receipt, and the same record often appears across multiple materials. When materials enter document parsing and chunking, procurement applications serve as the baseline for product name and quantity, usage registrations serve as the source for date and user information, inventory ledgers serve as the basis for balances and version continuity, and multilingual materials also need to preserve the correspondence between terminology and source text after chunking.
Configuration and implementation checks
This step is executed by parsing and training tasks (server-side tasks), outputting data rows and index statuses (parsed, indexing, indexed). The unified terminology glossary and comparison draft are then completed by retrieval and subsequent processing.
| Object | Check | Verification action |
|---|---|---|
| Procurement application materials | Whether product name, specification, quantity, date, and signature correspond item by item with the inventory ledger | Check the product name, quantity, and date fields of the procurement application against the inventory ledger, and inspect the signature field; the data row status should be parsed, indexing, or indexed |
| Usage registration materials | Whether the user, usage date, quantity, and receipt are consistent with the inventory ledger | Compare the name, quantity, and date of the usage registration with the inventory ledger, confirm the receipt is complete; separately verify the file record and the data row identity |
| Inventory ledger materials | Whether the version is synchronized, and whether the balance and the procurement application and usage registration can be traced back to the source text | Inspect the data rows covered by the rebuild task, confirm that only data with completed indexing is covered; check the chunk boundaries under the chunkSize and chunkSplitter settings |
| Multilingual terminology comparison draft | Whether the same term is written consistently throughout, and whether the translated name can be traced back to the source text | Sample three terms and check their spelling throughout; verify the processing mode corresponding to trainingType and the data row status transitions |
Three common errors
- Symptom: The name and quantity of the same batch of hazardous chemicals do not match between the procurement application and the inventory ledger. Cause: Multilingual material terminology is not unified, and after chunking each material is stored independently, lacking the constraint of a comparison draft.
- Symptom: The usage registration lacks a signature or receipt, but the ledger shows it as already balanced. Cause: After material parsing is stored in the database, it is not jointly verified with the signature receipt field, and the version enters subsequent processing before synchronization.
- Symptom: Newly uploaded materials cannot be found, and failed training leaves empty data rows. Cause: Data pending indexing does not participate in recall; successful parsing does not equal retrieval hits, so the index status must be checked.
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; the statuses should be parsed, indexing, and indexed, and the rebuild should only cover data with completed indexing.
Business test: Sample three terms and check whether their spelling is consistent throughout and whether the translated names can be traced back to the source text.
- Open the data rows to check the status, confirm it is consistent with the interface display, and that the file record and the data row identity are distinguishable.
- Inspect the data rows covered by the rebuild task, confirm that it targets only data with completed indexing, and that data pending indexing does not participate in the rebuild.