Data for this category
This category covers all documents and business data related to the end-to-end medical insurance fund supervision and audit process. Daily output documents include audit notices, inspection work drafts, suspicious point lists, interview records, handling decisions, and rectification reports. Data sources include medical insurance intelligent monitoring systems, fund supervision rule bases, medical insurance settlement data, drug and consumable traceability codes, and report clues. The update frequency is once per month. Each document includes fields such as business subject information, description of suspected violations, corresponding data traceability identifiers, and handling suggestions. Some documents link to multi-dimensional medical insurance settlement and traceability data, and the audit process requires combining multi-source information to complete.
Constraints Imposed on Knowledge Base Retrieval and Recall
Multi-source and heterogeneous data sources require the retrieval module to support structured settlement data, unstructured documents, and rule texts, and adapt to matching rules for different fields. The monthly update frequency means the knowledge base must be configured with regular synchronization tasks to ensure the timeliness of supervision rules and settlement data. The structure where a single document links to multiple traceability identifiers requires retrieval to support multi-field joint matching, accurately associating suspicious points with corresponding business data. The rigor of the audit process requires that recalled results include clear data sources and matching basis, to support the compliance of subsequent document drafting.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.72–0.85 | Matches the semantic similarity between audit suspicious point descriptions and business data. The interval covers the matching accuracy of supervision rules and suspicious points, and is calibrated based on actual tests |
limit | Top 8 entries | Number of associated documents returned per retrieval, adapts to the need for referencing multiple compliant documents and historical cases when drafting audit documents |
searchMode | hybrid | Hybrid retrieval mode, balances the recall effect of structured medical insurance settlement data and unstructured documents, adapts to multi-source data characteristics |
usingReRank | Enabled | Re-ranks initial recall results, raises the sorting priority of documents highly relevant to audit suspicious points, and improves the reference quality for document drafting |
datasetSearchUsingExtensionQuery | Enabled | Expands relevant supervision rules and settlement data keywords based on user queries, covering more comprehensive associated business information |
collectionIds | Pass the collection ID of the corresponding medical insurance fund supervision knowledge base | Limits the retrieval scope to the exclusive business knowledge base, avoiding mixing irrelevant non-medical insurance supervision data |
Three Common Issues
- A Query read timeout error occurs when calling the
/api/core/dataset/collection/create/apiColinterface. The cause is that the retrieval scope is not limited to the exclusive medical insurance fund supervision knowledge base, causing the system to load excessive redundant data that exceeds the interface processing threshold. - No search results are returned after passing in the knowledge base ID externally. The cause is that the
collectionIdsparameter is configured incorrectly, or the passed collection ID is not associated with business data related to medical insurance fund supervision. - No available knowledge base ID can be obtained in the workflow. The cause is that the exclusive medical insurance fund supervision knowledge base has not been created and synchronized on the FastGPT platform, or the ID information of the created collection has not been correctly extracted.
How to Verify Successful Configuration
- Enter the knowledge base management page of FastGPT v4.17.0, check the configuration items of the corresponding collection, and confirm that parameters such as
similarityandlimithave been configured according to the preset values. - Initiate a simulated audit suspicious point query, check whether the number and sorting of returned results meet expectations, and verify that the configurations of
usingReRankandsearchModetake effect. - Call the external API with the correct
collectionIdsparameter, verify that relevant medical insurance fund supervision documents can be returned, and confirm that the interface calling logic is correct. - View the knowledge base synchronization logs, confirm that the monthly update task has been executed normally, and no abnormalities have occurred in the data source.
Note: The configuration values provided here are common starting points, and should be evaluated against the reader's own specific 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.