What this category of data looks like
Core business documents for this category include audit notices, inspection work papers, suspicious item lists, interview records, handling decision documents, and rectification reports. Data sources cover medical insurance intelligent monitoring systems, fund supervision rule bases, medical insurance settlement data, drug and consumable traceability codes, and complaint clues. Data updates occur once per month. Document types include both structured forms and unstructured transcripts. Fields cover audit object information, descriptions of non-compliant behaviors, basis entries for handling, rectification requirements, and amount-related identifiers, among other content.
What constraints do these characteristics impose on the knowledge base retrieval and recall link?
The data characteristics of this category create multiple constraints for the retrieval and recall process. Dispersed multi-source data requires configuring multi-dataset associated retrieval to ensure full coverage of compliant policy and business data. The monthly update rhythm requires a matching knowledge base synchronization cycle to prevent data lag from impacting audit timeliness. Mixed-format document types need logic that balances structured field matching and unstructured semantic retrieval. High-risk supervision scenarios require strict control over the precision of recall results to avoid irrelevant content interfering with suspicious item screening and handling opinion formation.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the precision requirements of medical insurance audit policy texts, filters low-relevance redundant content |
limit | Top 10 entries | Controls the number of single-recall results, adapts to the scenario of auditors quickly reviewing content, avoids information overload |
searchMode | hybrid | Balances structured field matching and unstructured semantic retrieval, adapts to mixed-format documents in this category |
usingReRank | Enabled | Performs secondary sorting on recall results, raises the priority of highly relevant policy basis, meets the rigor requirements of audit scenarios |
datasetSearchUsingExtensionQuery | Enabled | Automatically expands search terms to cover synonymous expressions in policy texts, improves retrieval coverage |
collectionIds | Bound to fund supervision rule base and medical insurance settlement suspicious item dataset | Corresponds to the exclusive data source of this category, ensures that recall content only involves policies and documents related to medical insurance fund supervision |
Three Common Misconfigurations
- The interface sets
usingReRankto Enabled, but the re-ranking field shows false in retrieval return results. This occurs because the re-ranking model is not correctly associated with the current knowledge base retrieval node, or the input format of the re-ranking model does not match the recall results. - No optional values appear when referencing knowledge base variables in a workflow. This occurs because no variable corresponding to the knowledge base ID is created in the global variable configuration, or the variable's scope does not cover the current workflow node.
- The "Reference Variable" dropdown menu is empty when dynamically specifying a knowledge base. This occurs because the variable type is not set to knowledge base ID type, or the variable is not bound to the knowledge base resource pool of the current project.
How to Verify Successful Configuration
- Manually enter a question related to medical insurance audit suspicious items, check whether the
similarityscores of the retrieval return results fall within the configured range. - Check whether the number of return results from the retrieval node matches the number specified in the
limitconfiguration. - Enable debug mode in FastGPT v4.17.0, check whether the retrieval log shows that the enabled status of
usingReRankis true, and that the sorted results have the expected order. - Try using dynamic variables to replace the knowledge base selection option, confirm that the dropdown menu can load the configured variable options.
The values provided here are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.