What this category's data looks like
Undergraduate program establishment and dynamic adjustment involves three types of materials: program establishment application forms, talent cultivation plans, and expert review opinions. Each type contains fields such as program name, code, degree-granting category, length of study, course credits, number of faculty, review date, and signature. In the multilingual materials and terminology processing scenario, the same program name appears as the standard Chinese name in the application form, may include an English translated name in the cultivation plan, and may also appear as an abbreviation or former name in the expert review opinions. The terminology table must use the approved name in the application form as the baseline and trace each entry back to its original source. Materials are updated according to the annual application and adjustment cycle, versions iterate frequently, field formats are inconsistent, and date and signature fields often have gaps due to scanned copies or transcription differences. When the three types of materials enter the vector model and indexing, the application form plays the role of terminology baseline, the cultivation plan plays the role of expanding translated names and course terminology, and the expert review opinions play the role of terminology usage context and basis for changes, together forming a recallable terminology evidence set.
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). The unified terminology table and comparison draft are then completed by subsequent AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials).
| Object | Check | Verification action |
|---|---|---|
| Program establishment application form | Fields include program name, code, degree category, length of study, date, and signature; the name and quantity in the same record in the cultivation plan must be aligned item by item | Open a data row and check whether the index field contains multiple vector identifiers and whether the status is indexed, then compare the program name in the application form with the wording in the cultivation plan |
| Talent cultivation plan | Fields include course name, credits, class hours, and English translated name; the translated name must be traceable back to the standard Chinese name in the application form, and date and signature fields must not be empty | Check whether the translated names in the cultivation plan correspond one-to-one with the names in the application form, check whether the version number and review date are synchronized, and confirm that the rebuild task record covers this data scope |
| Expert review opinions | Fields include review conclusion, expert signature, review date, and receipt; the name, quantity, and date of the same record must be consistent across the three materials | Extract three terms and check whether their wording is consistent throughout the document, check whether signatures and receipts are complete, and check whether the vector identifiers in the index field are complete |
Three common errors
- Symptom: the same program name is written differently in the application form and the cultivation plan. Inspect the following: versions are not synchronized, and the terminology table does not trace each entry back to the original text based on the approved name.
- Symptom: expert review opinions lack signatures or receipts. Inspect the following: omissions in scanned copy transcription, fields are not completeness-checked, and dates and quantities do not match.
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the same group of data corresponds to multiple vector identifiers, retrieval does not merge back to the original data by identifier, and the vector score does not take the highest value.
How to verify the setup
Feature acceptance: extract one data row and check whether the index field records multiple vector identifiers and whether the status is indexed, then retrieve once to see whether data in the same group is merged and whether the score takes the highest value; after changing models, check the coverage scope of the rebuild task record.
Business test: extract three terms and check whether their wording is consistent throughout the document, confirm that translated names can be traced back to the original text, and that date and signature fields have no gaps.
- In the knowledge base settings, check whether the index model is consistent with the vector model actually used by the knowledge base, and check the tokens budget of the reference text corresponding to the
limitconfiguration items. - Open a data row and check whether the vector identifiers in the index field are complete, check whether the rebuild task record only covers data that has already been indexed, and separately record the number of entries in the citation array.