What this category's data looks like
This category's data covers three types of materials: registration records, course qualification certificates, and graduation application forms. Sources are provincial examination offices and main examining institutions, updated by examination session and graduation batch. Registration record fields include name, examination registration number, ID number, major code, course name, score, qualification date, signature, and receipt; course qualification certificate fields include course code, course name, qualification date, and seal; graduation application form fields include major name, number of courses, graduation conclusion, reviewer signature, and date. In multilingual materials and terminology processing, these three types of materials enter the vector model and indexing stage, respectively assuming the roles of raw data storage, vector recall, and index status recording.
Configuration and implementation checks
This stage 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 "unified terminology table and comparison draft" is then completed by subsequent AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials).
| Object | Check | Verification action |
|---|---|---|
| Registration record | The name, number of courses, and qualification date of the same record in the registration record and graduation application form must be consistent, and signatures and receipts must be complete | Open a data row and check whether the index field records multiple vector identifiers, and compare the names, quantities, dates, and signatures in the two materials |
| Course qualification certificate | The translated names of the course code and course name must be able to refer back to the original text, and the same term must be written consistently throughout | In the knowledge base settings, check that the index model is consistent with the vector model actually used by the knowledge base, and sample three terms to check their writing throughout |
| Graduation application form | The version must be synchronized with the registration record, and the reviewer signature and date must not be missing | Check the data scope covered by the rebuild task, and confirm that after changing the model, it is only triggered for data that has already been indexed |
| Data row | Multiple vector identifiers for one data row are all present; under the limit settings, the tokens of the quoted text are checked separately | Perform a retrieval to see whether data in the same group is merged, and check the status value against indexed and the rebuild task record |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: multiple vectors of the same group of raw data are not merged by identifier, and the vector score does not take the highest score.
- Symptom: the name, quantity, and date of the same record in the two materials do not match. Inspect the following: versions are not synchronized, and the update cadence of the registration record and graduation application form is inconsistent.
- Symptom: a data row has content but cannot be retrieved. Inspect the following: after changing the model, recall quality decreases, the rebuild task does not cover that data row, and the index field is missing vector identifiers.
How to verify the setup
This feature's acceptance: sample a data row to see whether the index field records multiple vector identifiers, then perform a retrieval to see whether data in the same group is merged, and check the status value against indexed and the rebuild task record.
Business test: sample three terms to check whether their writing is consistent throughout, and whether the translated names can refer back to the original text.
- In the knowledge base settings, check that the index model is consistent with the vector model actually used by the knowledge base, and check the tokens of the quoted text under
limit. - Check the data scope covered by the rebuild task, confirm that after changing the model, it is only triggered for data that has already been indexed, and check the number of array entries and the citation content budget separately.