What this category's data looks like
Enterprise new employee onboarding training materials mainly consist of onboarding training plans, course sign-in sheets, and job descriptions. Their sources are the human resources department, hiring departments, and training execution parties, and their update cadence changes with position adjustments, course revisions, and sign-in record archiving. When these materials enter the vector model and index feature, each plays a different role: the onboarding training plan provides the course framework and time nodes, the course sign-in sheet provides personnel participation and date evidence, and the job description provides responsibility and qualification requirement fields. Field characteristics are concentrated in names, quantities, dates, and signature columns; missing signatures or acknowledgments will prevent the material from serving as complete evidence. After materials are stored, the vector model generates vectors, and the index field records multiple vector identifiers. During retrieval, vectors are recalled first and then merged with the original data, and the highest score is taken for the same group of data.
Configuration and implementation checks
This part is executed by the index service and vector database, outputting vectors and index status (whether recall is possible and whether the same group of data 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).
| Object | Check | Verification action |
|---|---|---|
| Onboarding training plan | Whether the course names, quantities, and dates are consistent between the plan and the sign-in sheet | Open the data row to check the index field, confirm that multiple vector identifiers are complete, and after retrieval see whether the same group is merged |
| Course sign-in sheet | Whether the sign-in dates and signatures or acknowledgments correspond to the plan nodes | Check whether the data row status is indexed, and view the coverage of the rebuild task record |
| Job description | Whether the position name and responsibility fields are consistent with the course objects in the plan | In the knowledge base settings, check whether the index model is consistent with the vector model actually used |
| Data row index field | Whether multiple vector identifiers for one piece of data are all present | Open a data row to check and confirm that the index field records are complete |
Three common errors
- Symptom: The names, quantities, and dates of the same record do not match between the plan and the sign-in sheet. Inspect the following: The materials are maintained separately, and field alignment has not been done based on the uniqueness of the current version.
- Symptom: The old version of the job description is still retrieved, and citations are out of sync with the current version. Data rows whose indexing has not been completed do not participate in recall, so indexing must be completed first and then rebuilt.
- Symptom: The sign-in sheet lacks signatures or acknowledgments, so the material cannot serve as complete evidence. Inspect the following: The signature column was not checked during archiving.
How to verify the setup
This feature acceptance: sample one data row and look at the index field to confirm that multiple vector identifiers are all present; then retrieve once to see whether the same group is merged, with the highest score taken. When checking limit, verify the tokens of the cited text separately, record the number of array items separately, and do not compare the two for equality.
Business test: change one piece of content and see whether the version number and citations are updated synchronously.
- In the knowledge base settings, check whether the index model is consistent with the vector model actually used. After changing models, trigger a rebuild only for data that has already been indexed, and view the data range covered by the rebuild task record.
- Open a data row to check the index field, confirm that the status is indexed, multiple vector identifiers are complete, and after retrieval the same group of data is merged with the highest score taken.