What this category's data looks like
This page addresses the scenario of progress records and statistical verification for enterprise new employee onboarding training. The core materials are the onboarding training plan, course sign-in sheets, and job descriptions. The onboarding training plan provides course names, dates, responsible persons, and planned headcount; the course sign-in sheets record actual attendees, sign-in dates, and signed receipts; the job descriptions provide job titles, departments, and qualification requirements. The three types of materials are updated by batch or weekly, with dates, names, quantities, and signatures as common anchors. After the materials enter the knowledge base retrieval function, the onboarding training plan and job descriptions serve as sources for definitions and objects, while the course sign-in sheets serve as the record source. Together, the three support the generation of the progress table and the statistical definition explanation. Material versions must be synchronized with the weekly or monthly reporting cycle to avoid mismatches in names, quantities, and dates for the same record across two materials.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate snippets and similarity scores (after pattern, merge, rerank, filter, and limit trimming). The "progress table and statistical definition explanation" is completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Onboarding training plan | Check whether course names, planned dates, and responsible persons are consistent with the sign-in sheets and job descriptions | In the search test, switch searchMode and compare the number of candidate items and scores returned by the semantic, full-text, and hybrid modes |
| Course sign-in sheets | Check whether actual attendees, sign-in dates, and signed receipts match the plan and job descriptions | Test once each with synonym-rewritten questions to see whether candidate snippets cover the sign-in date, personnel names, and signature fields |
| Job descriptions | Check whether job titles, departments, and qualification requirements are consistent with the fields in the plan and sign-in sheets | Compare the order of the top few candidates before and after enabling usingReRank to confirm that the sorting of job title and department fields is stable |
| Progress table and statistical definition explanation | Check whether the record count and statistical count match and whether the definitions are written in the header | Adjust similarity and limit, observe changes in the number of candidate items and scores, and separately record the tokens and array length of the cited text |
Three common errors
- Phenomenon: The names, quantities, and dates of the same record do not match between the onboarding training plan and the course sign-in sheets. Cause: The two material versions are not synchronized or were updated in different batches, and the retrieval candidates do not cover the latest fields.
- Phenomenon: The job titles and department wording in the job descriptions and sign-in sheets are inconsistent. Cause: The materials lack a unified naming definition, the retrieved snippets come from different versions, and the similarity scores are close but the content conflicts.
- Phenomenon: The sign-in sheet lacks a signature or receipt, but the progress table still includes it in the total. Cause: The statistical definition does not write exceptions in the header, the retrieval candidates do not hit the signature field, and subsequent nodes directly use the record count.
How to verify the setup
This function's acceptance: test once each with synonym-rewritten questions, compare candidates and scores, then toggle rerank on and off to check the order of the top few; switch searchMode and usingReRank, and record changes in the number of candidates and scores.
Business test: use one week of records to verify the total, and check whether the definitions and exceptions are clearly written; check whether the record count and statistical count match and whether the definitions are written in the header.
- In the knowledge base search test, fix the question, switch between semantic, full-text, and hybrid modes respectively, record the number of candidates and scores, and confirm that the material fields are fully recalled.
- Compare the order of the top few candidates before and after enabling rerank, adjust
similarityandlimit, separately record the tokens and array length of the cited text, and then verify the totals in the progress table and statistical definition explanation.