What this category's data looks like
This page addresses the enterprise new employee onboarding training scenario. Materials include onboarding training plans, course sign-in sheets, and job descriptions. Fields cover course name, job title, date, quantity, signature, and receipt. Training plans are updated by batch, sign-in sheets are generated by session, and job descriptions are revised as the organization adjusts, so the three have different version rhythms. After multilingual materials enter the knowledge base, they are first split into data rows, then the vector model generates vectors and writes them into the index; the formation of the terminology glossary and the comparison draft depends on subsequent nodes. This feature only ensures that materials can be recalled and that data in the same group can be merged.
Configuration and implementation checks
This step is executed by the indexing service and the vector database, and outputs vectors and index status (whether recall is possible and whether data in the same group is merged). "Unified terminology glossary 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 |
|---|---|---|
| Onboarding training plan | Course names, job titles, dates, and signatures are written consistently across language versions and can be traced back to the source text | In the knowledge base settings, check the index model and confirm it is consistent with the vector model actually used by the knowledge base; open a data row and check whether the index field contains multiple vector identifiers |
| Course sign-in sheet | The name, quantity, and date of the same session match across the two materials, with no missing signature or receipt | Check the data scope covered by the rebuild task and confirm that after changing the model, it is triggered only for data that has already been indexed; check whether the data row status is indexed |
| Job description | The job title and terminology translation are consistent throughout, and the version is synchronized with the training plan | Sample a data row to inspect the index field, then search once to see whether the same group is merged; check that the tokens of the cited text and the number of array entries are recorded separately |
| Data row | Multiple vector identifiers for one piece of data are all present, and data in the same group can be merged by taking the highest score | Open a data row and check the index field; in the knowledge base settings, check the index model and confirm the splitting and budget settings corresponding to defaultToken and maxToken |
Three common errors
- Symptom: The course name, quantity, and date of the same record do not match between the training plan and the sign-in sheet. Inspect the following: The versions of the two materials are not synchronized.
- Symptom: After the job description is revised, the terminology translation is inconsistent throughout and cannot be traced back to the source text. Inspect the following: After changing the model, and old data is not covered.
- Symptom: The sign-in sheet lacks a signature or receipt, and duplicate entries appear in the search results. Inspect the following: compare the relevant source fields, document versions and downstream processing records.
How to verify the setup
Acceptance for this feature: sample a data row to inspect the index field, confirm that multiple vector identifiers are all present and the status is indexed; then search once to see whether the same group is merged, and check the data scope covered by the rebuild task records.
Business test: sample three terms and check whether their spelling is consistent throughout, and confirm that the translations can be traced back to the source text; record the number of citations by array length, and check quoteMaxToken separately as the citation content budget against tokens.
- In the knowledge base settings, check the index model and confirm it is consistent with the vector model actually used by the knowledge base; open a data row and check the index field and the splitting budget related to
defaultTokenandmaxToken. - Check the data scope covered by the rebuild task records and confirm that after changing the model, it is triggered only for data that has already been indexed; search once to check whether data in the same group is merged and whether the highest score is taken.