Knowledge Base Retrieval for International Student Admissions and Enrollment Records: Track Progress and Check Statistics

International Students: review retrieval for progress tracking. Check source versions, input mappings and the values received by downstream nodes.

What this category's data looks like

International student admissions and enrollment management involves three types of materials: admission brochures, admission letters, and visa application forms. Admission brochures contain the program, quota, application deadline, and signature field; admission letters contain the name, student ID, admission date, and issuer; visa application forms contain the passport number, entry date, and receipt. Materials are updated by semester batch, and field names, counts, dates, and signature status tend to drift between versions. When entering knowledge base retrieval and recall, admission brochures provide the program and quota definitions, admission letters provide enrollment identity and admission date, and visa application forms provide entry and receipt status. Together, the three support progress records and statistical verification, aligning the record count and statistical count under the header definitions.

Configuration and implementation checks

This step is performed by the knowledge base retrieval service (retrieval node and search test), outputting candidate chunks and similarity scores (after mode, merge, rerank, filter, and limit truncation). The progress table and statistical definition notes are completed by subsequent nodes or manually.

ObjectCheckVerification action
Admission brochureThe program, quota, application deadline, and signature field are consistent across the three materialsIn the knowledge base search test, switch searchMode, check the fields and dates of the brochure candidates, and record the number of candidates and scores
Admission letterThe name, student ID, admission date, and issuer signature are aligned across the two materialsTest once each with synonymous questions, compare the top few rankings before and after enabling usingReRank, and check the signature and receipt
Visa application formThe passport number, entry date, and receipt are synchronized across versionsCheck the original text chunks retained after similarity filtering, record the number of candidates and scores, and record the array length separately
Progress table and statistical definition notesThe record count and statistical count match, and the definitions are written in the header and include exceptionsVerify the totals weekly, check whether the header definitions and exceptions are clearly written, and check the cited text tokens

Three common errors

  • Symptom: The name and date for the same student do not match between the admission letter and the visa application form. Cause: The two material versions are not synchronized, and the retrieval candidates do not cover the latest version.
  • Symptom: The admission brochure quota and the progress table record count are inconsistent. Cause: After the brochure is updated, the old version is still recalled, and the statistical definitions are not written in the header.
  • Symptom: The visa application form lacks a signature or receipt, but the candidate chunk appears complete. Cause: After similarity filtering and citation limit truncation, the key signature field is cut off.

How to verify the setup

Acceptance for this feature: Test once each with synonymously rewritten questions, compare candidates and scores, then toggle rerank on and off to see the top few rankings; record the number of candidates and scores, and check similarity, limit, and searchMode.

Business test: Use one week of records to verify the totals, and see whether the definitions and exceptions are clearly written; record the number of citations separately by array length, and check the cited text tokens separately.

  • In the knowledge base search test, switch searchMode, compare semantic, full-text, and hybrid recall candidates, and record the number of candidates and scores.
  • Compare the order of the top few results before and after enabling usingReRank, and check the key original text chunks retained after similarity filtering and citation limit truncation.