What this category's data looks like
In the business of regular higher education institution admissions and enrollment filing, three types of materials—admission plans, filing rules, and admission rosters—together constitute the business basis. Admission plans include institution codes, major groups, planned numbers, subject selection requirements, and publication dates; filing rules include filing ratios, tie-breaking sorting, batch times, and approvals; admission rosters include candidate numbers, admitted majors, admission dates, and signed receipts. The three types of materials are updated by year and batch. The names, quantities, and dates of the same record often do not match across two materials, versions are not synchronized, or signed receipts are missing. When materials enter this function, admission plans and filing rules serve as rule-type bases, and admission rosters serve as result-type bases. They are split separately, written into data rows, and vector indexes are generated for cross-material basis query recall.
Configuration and implementation checks
This step is executed by the indexing service and vector database, outputting vectors and index status (whether recall is possible and whether data in the same group is merged). "Basis entries that can be traced back to the original text" is then completed by subsequent AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials).
| Object | Check | Verification action |
|---|---|---|
| Admission plan | Fields include institution code, major group, planned number, subject selection requirements, and publication date; the signature field and version number must be verifiable | Open one admission plan data row, check whether the index field records multiple vector identifiers, and confirm the status is indexed |
| Filing rules | Fields include filing ratio, tie-breaking sorting, batch time, and approval; the version must be current | In the knowledge base settings, check that the index model is consistent with the vector model actually used, and view the data range covered by the rebuild task record |
| Admission roster | Fields include candidate number, admitted major, admission date, and signed receipt; the name, quantity, and date of the same record must match the plan and rules | Retrieve one admission roster record, see whether data in the same group is merged after recall, and check the tokens budget of the cited text according to limit |
| Data row index field | Multiple vector identifiers for one piece of data are all present, and the array length is verifiable | Open one data row to check the index field, and confirm that the tokens of the cited text corresponding to limit and the number of array entries are recorded separately |
Three common errors
- Symptom: Duplicate entries appear in retrieval results. Inspect the following: The name, quantity, and date of the same record do not match across two materials, data in the same group is not merged, and taking the highest vector score does not take effect.
- Symptom: Recall quality drops after changing models. Inspect the following: Versions are not synchronized, and the rebuild task record only covers data whose indexing has been completed, not all materials.
- Symptom: A data row has content but cannot be retrieved. Inspect the following: Signatures or receipts are missing, the material has not completed indexing, the status has not reached indexed, and vectors have not been written to the vector database.
How to verify the setup
Acceptance of this function: Sample one data row and check the index field to confirm that multiple vector identifiers are all present; then retrieve once to see whether data in the same group is merged. For status values, only indexed and the rebuild task record are recognized.
Business test: Ask a cross-material basis question and see whether the answer provides traceable sources. The number of citations is the array length, and quoteMaxToken is the citation content budget; the two are not compared for equality.
- In the knowledge base settings, check that the index model is consistent with the vector model actually used by the knowledge base, and view the data range covered by the rebuild task record.
- Open one data row to check the index field, confirm that multiple vector identifiers are all present, check tokens of the cited text according to
limit, and record the number of array entries separately.