What this category's data looks like
This page addresses the enterprise new employee onboarding training scenario; materials include the onboarding training plan, course sign-in sheet, and job description. The onboarding training plan comes from the training department and is updated by batch; its fields include course name, class hours, instructor, and date. The course sign-in sheet comes from on-site sign-in and is updated by session; its fields include name, department, sign-in time, and signature. The job description comes from the hiring department and is updated by job version; its fields include job title, responsibilities, qualifications, and effective date. After these three types of materials enter the knowledge base, they respectively take on the roles of plan basis, attendance basis, and job responsibility basis. The same record's course name, quantity, and date in the plan and the sign-in sheet may not match; the job description version not being synchronized or the sign-in sheet missing a signature receipt will both affect whether the basis entry can point back to the original text.
Configuration and implementation checks
This step is executed by the index service and the vector database, outputting vectors and index status (whether recall is possible, whether data in the same group is merged). "Basis entries that can point back to the original text" 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 name, class hours, date, and signature column correspond item by item with the sign-in sheet | Open a plan data row and check whether the index field records multiple vector identifiers and whether the status is indexed |
| Course sign-in sheet | Whether name, department, sign-in time, signature or receipt are complete, and whether the same record's name and quantity are consistent with the plan | Retrieve the same course and check whether data in the same group is merged and whether the score takes the highest score |
| Job description | Whether the job title, responsibilities, qualifications, and effective date are the current version, and whether they are consistent with the version referenced by the training plan | Check the data range covered by the rebuild task record, and confirm whether data that has completed indexing after a model change triggers a rebuild |
| Data row | Whether the index model is consistent with the vector model actually used by the knowledge base, and whether limit are set according to the referenced content budget | Check the index model in the knowledge base settings, and open the data row to check the index field |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the same record's name, quantity, and date in the plan and the sign-in sheet do not match, and the vectors belong to different groups that are not merged.
- Symptom: recall quality drops after changing the model. Inspect the following: the version is not synchronized, and the rebuild task record only covers data that has completed indexing, not all materials.
- Symptom: a data row has content but cannot be retrieved. Inspect the following: a signature or receipt is missing, the index field does not record a vector identifier, and the status has not reached indexed.
How to verify the setup
This feature's acceptance: sample a data row to check the index field, then retrieve once to see whether the same group is merged; at the same time, check whether the status is indexed, and check the data range covered by the rebuild task record.
Business test: ask a basis question that spans materials, and see whether the answer gives a citation that can point back to the source.
- Check the index model in the knowledge base settings, and confirm it is consistent with the vector model actually used by the knowledge base.
- Open a data row to check the index field, confirm that multiple vector identifiers for one piece of data are all present, and check the data range covered by the rebuild task record.