What This Category's Data Looks Like
Data for this category comes from local public data open platforms, integrated big data platforms, departmental business databases, and electronic certificate repositories. Daily produced documents include public data resource catalogs, data open lists, data item descriptions, desensitization rules, interface documents, and update frequency descriptions. Data updates follow a monthly schedule. Core fields cover data item name, open scope, desensitization requirements, call interface address, update cycle, and authorization boundary. All content relates to structured or semi-structured public data assets.
Constraints Imposed on Knowledge Base Retrieval and Recall
The monthly update schedule requires knowledge base sync tasks to trigger on a fixed cycle. This prevents data from falling out of sync with actual open status. Structured and semi-structured data from multiple sources requires precise matching of data item-specific fields during retrieval. This avoids retrieving irrelevant generic content. Document types include interface documents and authorization boundary descriptions. Retrieval must support precise lookup of call permissions and desensitization rules, not just broad text descriptions. Data items have significant field differences. The number of single-round recall results must be limited. This avoids context overload that impacts model output.
Configuration Settings
These configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches structured fields of public data catalogs. This range filters low-relevance results while retaining recall content that aligns with authorization boundaries |
limit | Top 8 entries | Public data items are numerous. Limiting recalled entries avoids context overload, and supports the precise needs of policy basis retrieval |
searchMode | vector_only | Public data mostly consists of structured fields. Using only vector retrieval avoids ambiguity from keyword matching, and enables precise matching of data item characteristics |
usingReRank | Enabled | For structured data with multiple fields, re-ranking improves the ranking priority of highly relevant results, and supports the retrieval needs of policy basis |
datasetSearchUsingExtensionQuery | Disabled | Fields in public data catalogs are fixed. No extended queries are needed, which avoids introducing irrelevant generic content |
collectionIds | Fill in according to the actually bound public data catalog dataset | Precise specification of the retrieval scope prevents mixing in knowledge base content from other categories |
Three Common Misconfigurations
- No matching results are returned after the business system calls the retrieval interface: The cause is failure to associate the knowledge base added by the business system to the retrieval scope, or failure to correctly configure the
collectionIdsparameter to point to the corresponding dataset. - An error prompt is returned when adding a knowledge base: The cause is failure to upload documents that comply with public data catalog formats as required, or the sync task update cycle configuration exceeds the range allowed by the system.
- Retrieval result relevance decreases after long text input: The cause is failure to limit the token length of single input, or failure to enable reasonable retrieval and recall configurations, leading to context overload that interferes with model matching.
How to Verify Successful Configuration
- Manually upload a test public data item description document, trigger knowledge base sync, and confirm the sync status shows success.
- Input a query statement that includes a data item name, and verify that the retrieval results include the corresponding data open rules and interface descriptions.
- Adjust the value of the
similarityparameter, observe changes in the relevance of retrieval results, and confirm the configuration takes effect. - Call the retrieval interface of the business system, and verify that the number of returned results matches the configured
limitparameter.
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.