What the data for this category looks like
This category of data primarily originates from the national petition information system, local petition systems, the 12345 government service platform, grassroots governance platforms, and historical petition archives. The update frequency is daily. Daily documents include petition registration forms, acceptance notices, transfer and assignment letters, handling opinions, review and re-examination applications, resolution plans, and more. Fields include petition number, accepting entity, handling opinions, policy basis entries, appeal content, and others. Single documents are mostly a combination of structured business content and policy provisions.
What constraints these characteristics impose on the "source citation and traceability" link
The data sources cover multiple levels of government systems and are updated daily. This requires the citation and traceability link to support incremental synchronization and multi-source data correlation verification, to prevent the citation of old or non-official sources. Documents contain two types of content: structured business forms and policy provisions. It is necessary to accurately identify the policy basis fields to ensure that cited policy provisions correspond one-to-one with business handling opinions. Petition matter handling has extremely high requirements for traceability accuracy. Incorrect citations may lead to repeated petitions or review corrections. Therefore, the scope of cited sources and credibility levels must be strictly limited.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
| Single search limit | Top 3 entries | Petition policy basis materials are mostly official documents requiring precise matching; too many recall results will introduce irrelevant content |
| Similarity threshold | 0.75–0.85 | Match policy provisions and business documents for petition matters, to avoid low-relevance content being cited |
quoteMaxToken | 800–1200 token | Cover the complete content of petition handling opinions and policy basis materials, ensure citations are not truncated, and adapt to the configuration logic of FastGPT v4.17.0 |
responseData | Return only associated official petition and policy documents | Focus on the needs of policy and regulatory basis Q&A, filter non-essential business fields |
detail | Enabled | Display complete fields of cited documents, including core information such as petition number and policy basis entries |
Three common misconfigurations
- Phenomenon: Only two preset options (100 and 900) are available when configuring the citation limit, and intermediate values around 300 cannot be selected. Cause: The current version of the configuration interface does not provide an input box for custom ranges, only fixed preset options.
- Phenomenon: An error prompt appears when attempting to cite external HTTP returned content, and normal calls fail. Cause: The external data source call rules are not configured correctly, and valid access permissions or request header parameters are not added.
- Phenomenon: After setting the minimum relevance to 1, a large number of non-target citation content still appears in the response. Cause: The range of recalled data sources is not limited, and no exclusive petition knowledge base is associated, resulting in the recall of irrelevant content.
How to confirm the configuration is correctly set
- Submit a test question related to petition policies, and check whether the cited documents in the response come from the designated petition systems and policy sources.
- Review the field completeness of the cited content, and confirm whether core information such as petition number and policy basis entries is included.
- Adjust the configured citation token value, and verify whether the truncation of cited content in the response meets expectations.
- Check the synchronization status of the configured data sources, and confirm that daily incremental updates have been enabled normally.
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.