What data in this category looks like
Data for this category comes from government information disclosure platforms, official archive systems, document management systems, departmental business ledgers, and relevant standard documents. Daily documents include government information disclosure application forms, supplementary notice letters, third-party opinion solicitation letters, reply letters, service delivery receipts, administrative reconsideration response materials, and similar items. Data updates have no fixed cycle, and change alongside business processing progress or policy adjustments. Most documents combine structured and semi-structured formats, with fields including applicant information, application items, public attribute judgment basis, remedy approach descriptions, and others. There are no unified standard units, and corresponding rules must be matched according to business scenarios.
What constraints these characteristics impose on the knowledge base retrieval and recall link
Decentralized data sources require integrating heterogeneous data from multiple systems during retrieval, so cross-source retrieval adaptation logic must be configured. Diverse document types including forms, letters, reply letters and other structures require adjusting chunking and retrieval matching rules for different document formats. No fixed update cycle requires supporting on-demand incremental indexing to maintain data timeliness. Retrieval must accurately match the existence of information corresponding to the application items, public attributes, and content related to third-party rights and interests. Failure to do so may lead to processing delays or compliance risks.
Configuration settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Balance retrieval accuracy and coverage, avoid missing government information related to applications or introducing irrelevant content |
limit | Top 10 entries | Adapt to the scenario where a single application needs to retrieve multiple types of documents, cover necessary historical processing records and policy basis |
searchMode | vectorSearch | Prioritize semantic matching vector retrieval mode for semi-structured and unstructured government information documents |
usingReRank | Enabled | Improve the sorting priority of highly relevant results, ensure the information most matching the application items is returned first |
datasetSearchUsingExtensionQuery | Enabled | Automatically expand retrieval terms based on application items, cover synonyms and related expressions, avoid missing relevant government information |
collectionIds | Bind exclusive collections according to business departments | Limit the retrieval scope to the government information disclosure related data of the corresponding department, avoid interference from cross-department irrelevant data |
Three common misconfigurations
- Abnormally low number of retrieval results. The phenomenon is that fewer matching documents than the preset number are returned after entering an application item. The cause is that the
similaritythreshold is set too high, filtering out some relevant government information documents and failing to cover all required basis for the application. - Prefix field error during document import. The phenomenon is that the interface displays an error related to unrecognized
$schemafields. The cause is that non-business metadata in documents has not been cleaned, and the system cannot recognize custom schema fields. - Stuck index task. The phenomenon is that the index status remains at an intermediate progress for an extended period with no progress updates. The cause is that the
limitparameter is not set appropriately, resulting in too large a volume of document data loaded in a single batch, which triggers system timeout and interruption.
How to verify correct configuration
- Enter a typical government information application-based disclosure item, and confirm that the retrieved documents include historical processing materials and policy basis corresponding to the item.
- Check the running logs of the index task, and confirm that no
$schemarelated field errors appear, and the document import process completes normally. - View the document list in the knowledge base, and confirm that duplicate government information documents have been automatically cleaned or marked.
- Test the incremental index function, and confirm that newly added government information documents can be retrieved quickly, meeting the business requirements of irregular updates.
The values provided 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.