What the Data for This Category Looks Like
Service subjects for this category include natural resources department planning offices, planning institutes, local governments, and industry regulatory authorities. Daily documents cover territorial spatial master plans, detailed plans, special plans, planning implementation evaluation reports, and public opinion summaries. Data sources include the territorial spatial basic information platform, natural resources survey and monitoring data, population and economic statistics data, and remote sensing images. Data is updated annually. Documents contain fields such as spatial coordinates, planning indicators, and regional scope. Units mostly use hectares, square kilometers, ten thousand people, and similar units. Some documents are unstructured public opinion text.
What Constraints These Characteristics Impose on the Knowledge Base Retrieval and Recall Link
Annual updates create a large data volume. The data includes both structured indicators and unstructured text, so retrieval must balance precise matching and semantic completeness. Document structures vary significantly across types, so field extraction rules must be adapted for different documents to prevent accidental truncation of spatial coordinate content. Public opinion summary documents are mostly scattered colloquial texts, so chunking strategies need optimization to retain complete semantics. Documents generated through cross-department collaboration must support retrieval classified by source, to ensure coverage of all valid data sources and avoid missing cross-department planning coordination content.
Configuration Settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the indicator matching accuracy requirements of territorial spatial planning documents, prevents irrelevant results caused by overly low values, and avoids missing relevant planning content caused by overly high values |
limit | Top 8–12 results | Balances the comprehensiveness of retrieval results and context length, adapts to the large volume of multiple documents updated annually |
searchMode | hybrid | Combines keyword and semantic retrieval, adapts to document types that contain both structured indicators and unstructured opinion text |
usingReRank | Enabled | Optimizes the ranking of results related to spatial coordinates and planning scope, improves precise matching accuracy |
datasetSearchUsingExtensionQuery | Enabled | Expands the retrieval scope to cover associated policy documents based on the party building learning Q&A scenario |
collectionIds | Configured by planning type | Sets collection IDs classified by master plan, detailed plan, and other categories, to facilitate limiting retrieval scope by business scenario |
Three Common Misconfigurations
- Retrieval results mix non-planning documents. Check the document source field in interface previews or logs to identify this issue. The cause is failure to configure
collectionIdsby planning type, and failure to limit retrieval scope to only territorial spatial planning-related documents. - Insufficient retrieval results are returned. Verify the issue using the returned result count in the interface. The cause is an overly low
limitsetting, which fails to cover multiple planning documents updated annually. - Retrieval accuracy does not improve after enhancing index configuration. Compare retrieval results before and after enhancement to verify the issue. The cause is failure to optimize index rules for spatial coordinate fields, with only general text being enhanced.
How to Verify Proper Configuration
- Upload a special plan document, use a party building learning question that includes "territorial spatial master plan" for retrieval, and confirm returned results only include planning-related documents.
- Adjust the
similarityparameter to 0.8, use a question that includes "three zones and three lines" for retrieval, and confirm results contain accurate regional scope matching content. - After enabling
usingReRank, compare the ranking of retrieval results before and after enabling, and confirm ranking of spatial-related content has been optimized. - Check system logs to confirm
datasetSearchUsingExtensionQueryhas triggered normally, with no configuration error messages.
The values provided here are common starting points, and should be evaluated against the user’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.