Knowledge Base Retrieval for Vocational College Program Registration: Handle Questions and Collect Missing Information

Vocational Programs: review retrieval for question intake. Check source versions, input mappings and the values received by downstream nodes.

What this category's data looks like

Vocational college program setup filing materials usually include three types: the program setup filing form, talent demand justification, and faculty and equipment list. Fields cover program name, code, study duration, enrollment number, number of teachers, number of equipment units, justification date, and signature and seal. Material sources are college submissions, education administrative department acceptance receipts, and correction notices, and the update cadence advances according to the annual filing cycle and correction rounds. The program setup filing form plays the role of the main record, talent demand justification plays a supporting role, and the faculty and equipment list plays the role of condition verification. When the three types of materials enter the knowledge base retrieval and recall function, each is split and indexed as an independent segment for comparison of fields and dates during the acceptance stage.

Configuration and implementation checks

This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate segments and similarity scores (after mode, merge, rerank, filter, and upper-limit trimming). "Acceptance records and follow-up questions for completion" are completed by subsequent nodes or manually.

ObjectCheckVerification action
Program setup filing formThe program name, code, study duration, and enrollment number are consistent within the form, and the date and signature are completeIn the search test, test once each with synonymous questions, check the candidate segments and scores, and record the number of candidates
Talent demand justificationThe justification conclusion matches the program name, quantity, and date in the filing form, and the signature or receipt is not missingSwitch searchMode to hybrid and see whether the returned candidates include both semantic and full-text results
Faculty and equipment listThe number of teachers and number of equipment units match those filled in the filing form, and the version is the latestEnable usingReRank to compare the ranking of the top few items, and check the cited text after similarity filtering
Acceptance records and follow-up questions for completionMissing items are identified and followed up, and answers have material evidenceCheck whether the number of returned citations and the key original text segments are complete; the citation upper limit only checks cited text tokens

Three common errors

  • Phenomenon: The program name and quantity for the same record do not match between the filing form and the justification material. Cause: The two materials were filled in separately, and no field-level comparison was performed.
  • Phenomenon: The version of the faculty and equipment list was not synchronized, and old-version data was cited during acceptance. Cause: After correction, the new version was not re-ingested or did not overwrite the old segment.
  • Phenomenon: The filing form lacks a signature or receipt, but the acceptance record still proceeds as complete material. Cause: Missing-item identification did not trigger follow-up, and it went directly to the handler.

How to verify the setup

Acceptance of this function: test once each with synonymously rewritten questions, compare candidates and scores, then toggle rerank on and off to inspect the ranking of the top few items.

Business test: submit a question missing key information and see whether it first follows up before answering.

  • In the knowledge base search test, switch searchMode to semantic, full-text, and hybrid, and record the number of candidates and the similarity score respectively.
  • Run once with usingReRank enabled and once disabled, and compare the ranking of the top few items; the citation upper limit only checks cited text tokens, and the number of array items is recorded separately.