What this category's data looks like
This category's data covers ingredient names and specifications, retention sample times and batches, disinfection periods, and absence cause classification. These materials come from purchase orders, menus, retention sample records, disinfection records, and absence registers, and are updated daily or weekly. Ingredient names and specifications are the basis for reconciling purchasing and menus; retention sample times and batches link documents to physical items; disinfection periods cover activity rooms, tableware, and toys; absence cause classification is used for statistics and tracking. When the materials enter this feature, they are used as text or image descriptions to generate vector indexes, the original data is stored in business database data tables, and the index field records multiple vector identifiers to support subsequent retrieval and citation trimming.
Configuration and implementation checks
This step is executed by the indexing service and vector database, outputting vectors and index status (whether recall is possible, whether data in the same group is merged). The "version ledger and change records" are then completed by subsequent AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials).
| Object | Check | Verification action |
|---|---|---|
| Ingredient names and specifications | Whether the menu and purchasing categories are consistent | In the knowledge base settings, check that the indexing model matches the vector model actually used by the knowledge base, open a data row to check whether the index field contains multiple vector identifiers, and confirm that the data row status is indexed |
| Retention sample times and batches | Whether the documents and retention sample batches match | Retrieve the retention sample batch, check whether data in the same group is merged and whether the vector score takes the highest score, and check the data range covered by the rebuild task record |
| Disinfection periods | Whether the disinfection records are missing periods | Retrieve the disinfection periods, check whether the recall results cover activity rooms, tableware, and toys, and check the citation text budget corresponding to the index field and limit |
| Absence cause classification | Whether the classification is consistent with the register | Retrieve the absence causes, check whether data in the same group is merged, and check whether the rebuild task record is triggered only for data whose indexing has been completed |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the same group of original data corresponds to multiple vector identifiers, and after retrieval and recall they are not merged back into the original data by identifier, or the highest score is not taken during merging.
- Symptom: documents and retention sample batches do not match. Inspect the following: citation trimming takes the old version of the materials.
- Symptom: disinfection records are missing periods. Inspect the following: compare the relevant source fields, document versions and downstream processing records.
How to verify the setup
Feature acceptance: sample one data row and look at the index field to confirm that multiple vector identifiers are all present; then retrieve once more to see whether data in the same group is merged and whether the vector score takes the highest score, and check the rebuild task record and indexed status.
Business test: change one piece of content and see whether the version number and citations are updated synchronously; the number of citations is the array length, quoteMaxToken is the citation content budget, the two are not compared for equality, check the tokens of the citation text, and record the array count separately.
- In the knowledge base settings, check that the indexing model matches the vector model actually used by the knowledge base, and open a data row to check whether the index field contains multiple vector identifiers.
- Check the data range covered by the rebuild task record, confirm that after changing models it is triggered only for data whose indexing has been completed, and check
limit, and the citation text budget.