Vector Models and Indexes for Off-Campus Training Registration and Compliance: Find Supporting Evidence

Training Compliance: review vector indexes for evidence lookup. Check source versions, input mappings and the values received by downstream nodes.

What this category's data looks like

For off-campus training compliance filing and penalty response, standing materials include the school operating license, course filing materials, fee disclosure, and the corresponding receipts and signature pages. These materials are updated by semester or approval batch, and fields involve institution name, course name, number of class hours, fee amount, effective date, and signature. After entering this feature, the school operating license, course filing materials, and fee disclosure are each imported as independent datasets or groups, becoming retrievable source data; the vector model generates vector identifiers for each group of data, the source data and identifiers are stored in the business database, and the vectors are stored in the vector database. When material versions are out of sync or the name or date of the same record does not match, retrieval first recalls vectors, then merges back to the source data by identifier, taking the highest score for the same group of data, providing candidate support for subsequent reference back to the original text.

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). "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).

ObjectCheckVerification action
School operating licenseWhether the institution name, license number, validity start and end, signature and receipt are consistent with the course filing materialsOpen the data row where the school operating license is located, check whether the multiple vector identifiers in the index field are complete, and confirm the status is indexed
Course filing materialsWhether the course name, number of class hours, filing date, and signature match the fee disclosureIn the knowledge base settings, check whether the indexing model is consistent with the vector model actually used, then search the course name to see whether the same group is merged
Fee disclosureWhether the fee items, amount, disclosure date, and version are synchronized with the school operating license and course filing materialsCheck the data scope covered by the rebuild task, confirm that after changing the model only data that has completed indexing is triggered, and record defaultToken and maxToken
Data rowThe content exists and the index field is complete, and it can be recalledSample a data row to check the index field, then search once to see whether data in the same group is merged

Three common errors

  • Symptom: Duplicate entries appear in retrieval results. Inspect the following: The name or date of the same record in the school operating license and course filing materials does not match, and the vector identifiers are not merged into the same group of data.
  • Symptom: Recall quality drops after changing the model. Inspect the following: The rebuild task only applies to data that has completed indexing; data rows that have not completed indexing do not participate in recall, so indexing must be completed before rebuilding.
  • Symptom: A data row has content but cannot be retrieved. Inspect the following: The fee disclosure lacks a signature or receipt, the index field is missing vector identifiers, or the indexing model is inconsistent with the vector model actually used by the knowledge base.

How to verify the setup

Acceptance for this feature: sample a data row to check the index field, confirm that multiple vector identifiers are present and the status is indexed; then search once to see whether data in the same group is merged and whether the highest score is taken.

Business test: ask a cross-material basis question, for example whether the number of class hours in the course filing materials is consistent with the fee disclosure, and see whether the answer gives a traceable source.

  • In the knowledge base settings, check the indexing model to ensure it is consistent with the vector model actually used by the knowledge base, and record defaultToken and maxToken.
  • Open a data row to check the index field, confirm that multiple vector identifiers are present, and view the data scope covered by the rebuild task record.