Policy and Regulatory Basis Q&A for Fiscal Expenditure Performance Evaluation: Citation Sources and Traceability

Reference only; verify against the target FastGPT release and deployment documentation.

What this category of data looks like

This category of data comes from the unified budget management system, project application materials, fund payment data, business system ledgers, and satisfaction survey results. Commonly processed documents include performance target forms, self-evaluation forms, performance evaluation reports, supporting material lists, and rectification reports. The data update cycle is annual. Most document structures include project numbers, performance indicators, fund details, and supporting material association fields. There is no unified fixed unit, and fields vary slightly depending on the project type.

What constraints do these characteristics impose on the citation sources and traceability link

The data sources for this category are scattered and linked to specific projects, requiring the traceability link to accurately bind project numbers to corresponding policies and business data, to avoid generalized citations. The annual update feature requires regular refreshing of the knowledge base’s policy and business data to ensure cited content is not outdated. Multiple document types exist, so it is necessary to distinguish between performance evaluation documents and policy basis documents to ensure accurate identification of source categories during traceability. The strong association between supporting materials and performance targets requires traceability results to be linked to specific fund expenditure links and bound to actual business policy entries. Document formats submitted by different entities vary, so the traceability link must support metadata extraction from multiple document formats to ensure consistency of citation sources.

How to set the configurations

Three common configuration mistakes

• Issue: Officially published chat pages do not display citation sources, and only debug pages show this information. Cause: The responseData configuration item is not enabled, or the response data metadata carrying rules are not correctly configured.
• Issue: Setting the citation limit to 2000 causes single-answer tokens to exceed the threshold and be truncated. Cause: quoteMaxToken is not adjusted based on the average length of this category of documents; an excessively high citation limit causes aggregated content to exceed the model’s context window limit.
• Issue: Traceability results do not link to specific project information, and only display policy names. Cause: The detail configuration item is not enabled, and associated fields such as project numbers and performance indicators in the document are not extracted, leading to generalized traceability information.

How to confirm the configuration is correct

• Navigate to the officially published chat page, initiate policy inquiries related to fiscal expenditure performance evaluation, and check if citation source entries are displayed at the end of responses.
• View the quoteMaxToken value on the knowledge base configuration page, confirm it matches the average length range of this category of documents.
• Trigger debug mode, view the complete response data, confirm that the responseData field contains metadata from source documents.
• Submit updated policy documents, verify that traceability results can correctly link to the latest policy versions.

The values provided here are common starting points, and should be evaluated against individual deployment 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.