What this category of data looks like
Coal mining production policy and regulatory basis data is primarily sourced from public normative documents, industry standards, and local regulatory rules issued by the State Energy Administration, National Mine Safety Supervision Administration, and Ministry of Emergency Management. Data updates are triggered by the release of new policies, with no fixed cycle. Each individual document includes document number, issuing body, release date, applicable mine level (underground/open-pit), core clause text, regulatory requirements, and some documents include attached tables. Content is split by clause hierarchy, with no unified unit field. Only clause numbers and page numbers are used as positioning identifiers.
What constraints these characteristics impose on the citation sources and traceability link
Coal mining production policy documents have multiple tiers and dispersed sources. Traceability links require precise matching of document issuing bodies and document numbers, to avoid confusion between cross-level regulatory documents. Some policy documents include attachment tables, so support for associating attachments as complete traceability basis is required. Policy documents have no unified unit field, so clause numbers and page numbers must be used as positioning anchors. Precise traceability cannot be completed only through content fragments. Policy updates have no fixed cycle, so the knowledge base’s document versions must be synchronized regularly to ensure traceability results align with the latest valid policies.
How to configure the setup
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 token | Matches the average length of single clauses in coal mining policy documents, to avoid truncation of core traceability information |
responseData | Enable traceability field return | Ensures returned citation data includes key positioning information such as document numbers and issuing bodies |
detail | Full mode | Retains clause numbers and page numbers from original documents, supporting precise traceability |
| Single search limit | Top 5 results | Coal mining policy documents are relatively concentrated in quantity, to avoid redundant results interfering with matching |
| Traceability association of attachments | Enable | Covers scenarios of policy documents with attached tables |
Three common configuration errors
- Phenomenon: "No permission to operate this conversation record" is displayed at the end of the response, and citation traceability information cannot be viewed. Cause: The citation permission switch for the knowledge base is not configured, or the current conversation role is not granted permission to view traceability information.
- Phenomenon: Citation results only return text fragments, and complete document numbers and issuing bodies cannot be matched. Cause: The
detailfull mode is not enabled, only content fragments are returned without fields required for positioning. - Phenomenon: The number of citations returned by knowledge base search exceeds the expected limit. Cause: The single search limit is not set, or the configured limit value is higher than the actual available number of coal mining policy documents, leading to redundant results being included.
How to confirm the configuration is complete
- Launch a test query about coal mine safety policies, and check whether the citation information at the end of the response includes positioning fields such as document numbers and issuing bodies.
- Adjust the value of
quoteMaxToken, and test whether citation content of different lengths is returned completely without truncation. - After importing and exporting the workflow, check that traceability-related configuration items are fully retained with no missing parameters.
- Switch between different conversation roles, and verify that citation traceability information can be viewed normally with no permission errors.
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.