What this category's data looks like
In the graduate admissions re-examination and admission scenario, the re-examination plan, the candidate score sheet, and the transfer/adjustment list are the three core types of materials. The re-examination plan contains assessment dimensions, weights, time and location, and a signature field; the candidate score sheet contains the candidate number, subject scores, converted scores, and a review signature; the transfer/adjustment list contains the transfer batch, quota, proposed admission status, and receipt. The materials mostly come from departments and the graduate admissions office, and are revised multiple times before the re-examination, after the re-examination, and during the transfer stage. After entering this feature, the re-examination plan is used as rule text to generate a text index, while the score sheet and the transfer/adjustment list enter the knowledge base as structured records; each record retains the original data in the business database and records multiple vector identifiers. The field characteristics are the coexistence of dates and signatures, the repeated appearance of the same record across materials, and an update rhythm that changes as batches progress.
Configuration and implementation checks
This part is executed by the indexing service and the 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). The indexing model must be consistent with the vector model actually used by the knowledge base, and the index field of a data row must fully record the multiple vector identifiers of that piece of data.
| Object | Check | Verification action |
|---|---|---|
| Re-examination plan | The fields contain assessment dimensions, weights, dates, and a signature field; the name, quantity, and date of the same record in two materials must match | Check the indexing model in the knowledge base settings, and open a re-examination plan data row to check the index field |
| Candidate score sheet | Contains the candidate number, subject scores, converted scores, and review signature; the scores and dates of the same candidate in two materials must be consistent | Retrieve a candidate record, confirm whether data in the same group is merged, and view the data range covered by the rebuild task record |
| Transfer/adjustment list | Contains the transfer batch, quota, proposed admission status, and receipt; it must be marked when the version is not synchronized or the signed receipt is missing | Check whether the data row status is indexed, and check that the tokens and array count of the cited text are recorded separately |
| Indexing model and rebuild | The indexing model is consistent with the vector model actually used by the knowledge base; after changing the model, trigger a rebuild only for data that has already been indexed | View the data range covered by the rebuild task record, and confirm that recall quality has not declined after changing the model |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the name, quantity, and date of the same record in two materials do not match, and multiple vector identifiers are not merged into the same group of original data.
- Symptom: recall quality declines after changing the model. Inspect the following: the indexing model is inconsistent with the vector model actually used by the knowledge base, and the rebuild task record does not cover all data that has already been indexed.
- Symptom: a data row has content but cannot be retrieved. Inspect the following: the version is not synchronized, the old version is still referenced, or the transfer/adjustment list lacks a signature and receipt, so citation trimming cannot obtain supporting materials.
How to verify the setup
This feature's acceptance check: sample a data row to see whether the index field is complete, then retrieve once to see whether data in the same group is merged and whether the highest score is taken; check whether the values of defaultToken and maxToken match the indexing model, and confirm that the status is indexed.
Business test: change one piece of content, see whether the version number and citations are updated synchronously, and confirm that the status is indexed and consistent with the rebuild task record.
- Check the indexing model in the knowledge base settings, and open a data row to check whether the index field records multiple vector identifiers.
- View the data range covered by the rebuild task record, and confirm that after changing the model, a rebuild is triggered only for data that has already been indexed.