What the Data for This Category Looks Like
Data sources for this category include government websites, document management systems, government information disclosure directories, policy document repositories, and press conference materials. The update frequency is daily. Daily documents include policy documents, policy interpretation drafts, public disclosure directories, information release approval forms, press releases, and website column update records. Data fields cover document title, document number, release date, public disclosure attribute, secrecy level, approval status, associated release columns, and more. The length of individual documents ranges from hundreds to tens of thousands of words.
What Constraints Do These Characteristics Impose on the "Knowledge Base Retrieval and Recall" Link
Daily updated data sources require the knowledge base sync frequency to align with the update rhythm, to avoid recalling outdated or unapproved policy content. The wide variation in document lengths across types requires the retrieval chunking strategy to adapt to different text lengths, to avoid truncating short approval forms or failed chunking of long policy documents. Sensitive fields such as public disclosure attribute and secrecy level require retrieval results to be associated with corresponding approval node data, to prevent accidental recall of sensitive information. The field for associated release columns requires retrieval results to match the category rules of the target release channel, to ensure recalled content complies with the compliance requirements of the release scenario.
How to Configure the Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Must filter low-relevance policy interpretations or approval forms, while retaining associated content marked with secrecy attributes; interval calibrated based on actual testing |
limit | Top 8 entries | Adapts to the batch review needs of government public disclosure documents, avoiding excessive results exceeding the processing limit of file review |
searchMode | hybrid | Balances keyword matching and semantic recall, adapting to the mixed retrieval needs of structured fields such as policy document numbers and titles, as well as content text |
usingReRank | Enabled | Performs secondary ranking on initial recall results, filtering low-quality matching content caused by semantic ambiguity |
datasetSearchUsingExtensionQuery | Enabled | Generates extended query terms based on user questions, covering scenarios such as aliases and abbreviations of policy documents, improving recall coverage |
collectionIds | Bind to government public disclosure-specific datasets | Only retrieves specific data such as policy documents and approval forms for this category, avoiding mixing irrelevant content from other categories |
Three Common Mistakes
- Phenomenon: Retrieval results include documents with unmatched associated IDs, making it impossible to link to the corresponding approval process. Cause:
collectionIdsis not configured correctly, and the bound dataset does not include government public disclosure documents with associated ID fields. - Phenomenon: The knowledge base retrieval link is not triggered when the workflow runs, and a generic reply is returned directly. Cause: The knowledge base retrieval step was not added correctly to the workflow node, or
datasetSearchUsingExtensionQuerywas not enabled. - Phenomenon: The number of retrieval results does not match the configured
limit, with too many or too few results. Cause: Thesimilaritythreshold was not set correctly, resulting in abnormal filtering rules or the recall limit not taking effect as expected.
How to Confirm the Configuration Is Correct
- Enter the knowledge base management page of FastGPT v4.17.0, check whether the bound dataset is a government public disclosure-specific collection, and confirm that the
collectionIdsconfiguration is correct. - Submit a test question containing a policy document number, check whether the retrieval results prioritize returning the corresponding policy document, and verify that the
searchModeandsimilarityconfigurations are effective. - Run the workflow test case, check whether the log contains the retrieval results of the knowledge base, and confirm that
datasetSearchUsingExtensionQueryand the retrieval node have been added correctly. - Trigger a daily sync task, check whether the knowledge base update log matches the update frequency of the data source, and verify that the sync configuration meets the requirements.
The values listed here are common starting points and should be evaluated 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.