What Does the Data for This Category Look Like?
Data sources for this category include the national land spatial planning one-map, project approval and supervision platform, current land use database, and ecological protection redline data. Daily documents cover land pre-examination and site selection opinions, construction land planning permits, construction engineering planning permits, redline diagrams, planning conditions, and similar materials. Data update frequency is irregular, synchronized with project approval progress or adjustments to national land spatial planning. Most documents are structured approval forms, spatial vector files, and text descriptions. Core fields include project site selection, land use scale, planning control requirements, and similar content. There are no unified fixed units, adjusted based on the specific attributes of each project.
What Constraints Do These Characteristics Impose on Knowledge Base Retrieval and Recall?
Data sources for this category are scattered and include multiple formats such as spatial vector files and approval forms. Retrieval must be limited to specified approval documents and party building learning materials to avoid interference from irrelevant data. Data updates are irregular, so incremental synchronous recall must be supported to ensure retrieval results include the latest control requirements. Core fields involve sensitive content such as project site selection and redline conflicts. Retrieval results must accurately match project attributes to avoid situations where control rules do not align with actual projects. Additionally, the party building learning Q&A scenario requires retrieval results to be linked with policy interpretations and approval processes. Additional party building-related learning materials must be associated to further limit the retrieval scope and improve the targeting of learning content.
Configuration Recommendations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the precision requirements of land pre-examination and planning permit control rules. Avoids irrelevant results caused by too low a threshold, and prevents missing compliant matches caused by too high a threshold. |
limit | Top 8 entries | Balances the context length of party building learning Q&A (single context typically does not exceed 2000 characters) and information completeness. Too many single recall entries will exceed the model's context limit. |
searchMode | Exact match | Adapts to the retrieval needs of structured approval documents and spatial data, avoiding redline conflict misjudgments caused by fuzzy matching. |
usingReRank | Enabled | Re-ranks initial recall results according to party building policies and approval rules, ensuring core control requirements are displayed first. |
collectionIds | Configured according to specified approval document libraries and party building learning libraries | Limits the retrieval scope, only calling documents related to land pre-examination and planning permits and party building learning materials (total data volume does not exceed 500 MB). |
datasetSearchUsingExtensionQuery | Enabled | Automatically optimizes questions related to party building learning, supplements context information of approval processes, and improves retrieval accuracy. |
Three Common Misconfigurations
- Retrieval results only return partial table entries, failing to cover the full preset number of entries. The cause is incorrect configuration of the
limitparameter or failure to enable full table parsing, resulting in insufficient recall entries.
Reference only; verify against the target FastGPT release and deployment documentation.
- Multiple file partitions are missing in backed up and imported knowledge bases. The cause is that the dataset was exported and imported without following the original instance's dataset structure, only exported as a single CSV file, failing to retain the dataset's associated relationships.
How to Confirm the Configuration Is Correct
- Submit a test question that includes specific project site selection and control requirements, and check whether the retrieval results include the core fields of the corresponding documents.
- View the retrieval log to confirm that the source library of the recall results matches the range configured in
collectionIds. - Trigger incremental data synchronization, and verify whether newly uploaded approval documents can be normally recalled.
- Adjust the
similarityparameter range, observe changes in the matching accuracy of retrieval results, and confirm that they meet business requirements.
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.