Internal Office and Employee Services for Territorial Spatial Planning Compilation and Implementation Assessment: Knowledge Base Retrieval and Recall

What Data for This Category Looks Like

Daily documents for this category include territorial spatial master plans, detailed plans, special plans, planning implementation assessment reports, public opinion summaries, and similar materials. Data is primarily sourced from the territorial spatial basic information platform, natural resource survey and monitoring data, population and economic statistics data, and remote sensing image products. Data update cycles are typically annual. Most documents contain fields such as spatial coordinates, constraint indicators, and regional scope. Some documents include associated geospatial layer information. Data volume varies based on planning hierarchy and coverage scope.

Constraints on Knowledge Base Retrieval and Recall From These Data Characteristics

The annual update feature of the data requires retrieval and recall to support version verification, to avoid returning expired planning content. Multi-source heterogeneous data sources create differences in fields and formats. Retrieval logic must support mixed matching rules for spatial coordinates, statistical indicators, and text descriptions. Documents contain dedicated fields such as regional scope and constraint indicators. Retrieval must prioritize content directly related to the current planning compilation task’s region and control requirements. The increasing proportion of unstructured documents such as public opinion summaries requires optimizing recall accuracy after long-text chunking, to prevent key opinions from being missed.

How to Configure Settings

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

Three Common Configuration Errors

• Retrieval result count is far lower than expected, with only a small number of planning documents returned. The cause is failure to adjust the limit configuration item’s value. The default count is insufficient to cover recall needs for multi-source planning data.
• Retrieval results include expired old versions of planning content. The cause is failure to associate data version verification logic, and failure to filter historical documents created before annual updates.
• Associated documents containing spatial coordinates cannot be recalled. The cause is failure to select searchMode adapted to multi-source data, relying solely on pure vector retrieval and ignoring spatial keyword matching.

How to Confirm Proper Configuration

• Import a single typical planning document, initiate a test query containing a regional name and control indicator, and verify whether returned results include relevant fragments of that document.
• Adjust the value range of the similarity configuration item, compare recall results across different thresholds, and confirm that matching accuracy meets current task requirements.
• After enabling usingReRank, check whether re-ranked results prioritize content directly related to the current compilation region.
• Compare retrieval results with and without datasetSearchUsingExtensionQuery enabled, and confirm that expanded queries have improved recall rates for relevant content.

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.