What this category's data looks like
In the enrollment consultation conversion scenario at training institutions, materials are mainly of three types: enrollment brochures, consultation records, and trial class schedules. Enrollment brochures contain course names, class types, number of class hours, start dates, and seals; consultation records contain the consultant, intended course, communication date, and consultant signature; trial class schedules contain student names, trial class times, teaching instructors, and receipts. Materials are updated by enrollment batch and date, and the same record is mutually corroborated across brochures, consultation records, and trial schedules by name, quantity, and date. When entering the knowledge base retrieval and recall function, brochures and trial class schedules serve as semantic and full-text recall corpus, consultation records serve as associated materials for follow-up questions and review, and the retrieval service is only responsible for recalling candidate snippets and similarity scores, and does not generate statistical tables, business documents, or authorization results.
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 mode, merge, rerank, filter, and limit trimming). The "exception list and review conclusions" are completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Enrollment brochure | Whether the course name, class type, number of class hours, start date, and seal are consistent with the consultation records | In the knowledge base search test, use synonymous questions to retrieve, note the number of candidates and scores, and verify the course name, quantity, date, and seal in the original brochure text |
| Consultation record | Whether the consultant, intended course, communication date, and consultant signature match the trial class schedule | Switch searchMode to semantic, full-text, or hybrid, compare whether the candidates cover both the consultation record and the trial class schedule, and check the signature and receipt |
| Trial class schedule | Whether the student name, trial class time, teaching instructor, and receipt are synchronized with the brochure version | Compare the top few rankings before and after enabling usingReRank, and check whether the similarity score and candidate snippets retain the trial class time and instructor fields |
| Retrieval node and search test | Whether the candidate snippets and similarity scores have undergone mode, merge, rerank, filter, and limit trimming | Record the number of returned array items and the tokens of the cited text, and verify whether the key snippets of the original text are still retained after limit trimming |
Three common errors
- Symptom: The course name or date for the same student does not match between the consultation record and the trial class schedule. Cause: The two materials were updated in different batches, versions were not synchronized, and the retrieval recalled snippets from an old version.
- Symptom: The number of class hours stated in the enrollment brochure is inconsistent with the number promised by the consultant in the consultation record. Cause: After the brochure was revised, the consultation record was not backfilled, review signatures were missing, and candidate snippets could not corroborate each other.
- Symptom: The trial class schedule lacks a receipt or consultant signature, and the exception list cannot be traced back to the source materials. Cause: Receipts were not archived, retrieval candidates only covered the schedule, and supporting associated records were missing.
How to verify the setup
Acceptance of this function: Test once each with synonymously rewritten questions, compare the number of candidates and similarity scores, then toggle usingReRank to observe the top few rankings, and switch searchMode to observe whether the candidates include both semantic and full-text results.
Business test: Create an exception record, see whether it can be traced back to the source materials, and confirm that the names, quantities, dates, and signatures in the enrollment brochure, consultation record, and trial class schedule can be checked item by item.
- In the knowledge base search test, retrieve once each with synonymous questions, record the number of candidates and scores, and verify whether the returned snippets retain the key fields of the original text.
- Observe the top few rankings once with rerank enabled and once disabled, and check the tokens of the cited text and the number of array items to confirm that the source materials can still be traced after trimming.