What this category's data looks like
This page focuses on three types of materials for enterprise independent skill talent evaluation: enterprise evaluation filing forms, evaluation specifications, and candidate rosters. The filing form provides the evaluation entity, occupation name, and filing date; the evaluation specification gives application conditions, assessment methods, and scoring criteria; the roster records candidate names, ID numbers, applied occupations, and batches. The three types of materials are updated by batch, with fields mainly consisting of names, quantities, dates, signatures, and receipts. When entering knowledge base retrieval and recall, the filing form is used to confirm whether the evaluation business has a basis, the evaluation specification is used to answer questions about conditions and procedures, and the roster is used to verify personnel and batch information. Together, the three constitute a basis source that can be traced back to the original text.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate fragments and similarity scores (after mode, merge, rerank, filter, and limit trimming). The basis entries that can be traced back to the original text are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Enterprise evaluation filing form | Fields include evaluation entity, occupation name, filing date, and signature; must be compared item by item with the evaluation specification and roster | In the search test, use searchMode to switch between semantic, full-text, and hybrid, check the number of candidate entries and scores for the filing form, and verify whether names, dates, and signatures match |
| Evaluation specification | The version must be current; application conditions, assessment methods, and scoring criteria must be consistent with the filing form | Enable usingReRank to compare the order of the top few entries before and after reranking, and verify whether the specification version number and the occupation listed in the filing form are synchronized |
| Candidate roster | Candidate names, ID numbers, applied occupations, and batches must correspond to the filing form and evaluation specification | Record the number of candidate entries and scores, verify whether the number of people in the roster matches the planned number in the filing form, and check whether receipts and signatures are complete |
| Knowledge base retrieval service | Candidate fragments and similarity scores come from multi-route recall merge, rerank, filter, and limit trimming | Use similarity to observe the filtering effect, use limit to observe the number of cited entries, and separately record the cited text tokens and array count |
Three common errors
- Phenomenon: The same candidate name is written differently in the roster and the filing form. Cause: The name standard was not unified during data entry, so cross-material comparison fails.
- Phenomenon: The evaluation specification version is not synchronized, and old application conditions are still cited. Cause: The material update pace is inconsistent, and the current version has not been replaced.
- Phenomenon: The filing form lacks a signature or receipt, and the basis entry cannot be traced back to the original text. Cause: The material archiving step omitted it, and the retrieved fragments lack verifiable fields.
How to verify the setup
Acceptance of this feature: test once each with synonym-rewritten questions, compare candidates and scores, then toggle reranking on and off to see the order of the top few entries.
Business test: ask a cross-material basis question and see whether the answer provides a traceable source.
- In the search test, switch
searchMode, record the number of candidate entries and scores, and verify whether cross-material names, quantities, and dates are consistent. - Enable
usingReRankto compare the order of the top few entries, usesimilarityandlimitto check the filtering and trimming effects, and separately record the cited text tokens and array count.