What this category's data looks like
This category's data looks like this General college entrance examination registration qualification review involves three core types of materials: the registration form, household registration and student status proof, and materials for children accompanying migrating parents. The registration form contains the candidate's basic information, examination category, and signature field; the household registration and student status proof contains the household registration address, student status start and end dates, and the school seal; the materials for children accompanying migrating parents contain the residence permit, social insurance payment records, and proof of parents' employment. These materials are submitted intensively during the registration season, and multiple corrections may be made within the review window. The names, quantities, and dates of the same record often differ across different materials, and after version updates it is also common for the old version not to be synchronized, or for signatures or receipts to be missing. When the materials enter the knowledge base retrieval function, the registration form and household registration and student status proof serve as the main sources for semantic and full-text recall, while the materials for children accompanying migrating parents supplement the basis fragments for the residence and social insurance dimensions. Together, the three provide candidates for subsequent nodes or manual generation of basis entries 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 then completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| General college entrance examination registration form | Fields include candidate name, examination category, and signature field; the date must match the student status start and end dates in the household registration and student status proof, and the signature field must not be empty | In the search test, use "registration form signature field and student status start and end dates" as the query, switch searchMode to compare semantic and full-text candidates, and record the number of candidates and scores |
| Household registration and student status proof | Fields include household registration address, student status start and end dates, and school seal; the name and date of the same record in the two materials must be consistent, and the version must be current | Test once each with synonymous rewritten queries, check whether the candidates include both semantic and full-text results, and compare the ranking of the top few before and after enabling usingReRank |
| Materials for children accompanying migrating parents | Fields include residence permit number, social insurance payment records, and proof of parents' employment; the quantities and dates must match those in the registration form and household registration and student status proof, and missing receipts must be marked | Use "basis for social insurance and residence permit of children accompanying migrating parents" as the query, check whether the candidate fragments can be traced back to the original text, record the similarity score and the number of candidates, and after limit trimming check whether the key fragments are retained |
Three common errors
- Phenomenon: The same candidate's name is written differently in the registration form and the household registration and student status proof, and the dates also do not match. Cause: During material entry, field-level comparison was not performed for the same record, and the version was not synchronized.
- Phenomenon: The number of social insurance payments in the materials for children accompanying migrating parents is inconsistent with the number filled in the registration form, and signatures or receipts are missing. Cause: After correction, the old version was not replaced, and receipts were not archived.
- Phenomenon: The answer lacks key paragraphs, and although there are many candidates, the ones ranked at the front are all irrelevant. Cause: The retrieval mode does not match the material form, rerank was not enabled, or the similarity filter is too strict.
How to verify the setup
Acceptance of this function: test once each with synonymous rewritten queries, compare candidates and similarity scores, then toggle usingReRank on and off to observe the ranking of the top few; record the number of candidates and scores, and after limit trimming check the tokens of the cited text, recording the number of array items separately.
Business test: ask a cross-material basis question, see whether the answer gives a traceable source, and check whether the basis comes from the institution’s materials, whether the version is current, and whether the entries match the original text.
- In the knowledge base search test, switch
searchModeto compare the candidate results of semantic, full-text, and hybrid options, and confirm that it matches the material form. - Compare the order of the top few before and after enabling
usingReRank, and check whether the key fragments of the original text are retained aftersimilarityfiltering andlimittrimming.