Internal Office and Employee Services for Territorial Spatial Planning Compilation and Implementation Evaluation: Document Parsing and Chunking

Internal Office and Employee Services for Territorial Spatial Planning Compilation and Implementation Evaluation: Document Parsing and Chunking

What the data for this category looks like

The daily documents for this category include territorial spatial overall plans, detailed plans, special plans, planning implementation evaluation reports, public opinion summaries, and similar materials. Data mainly comes from the territorial spatial basic information platform, natural resource survey and monitoring data, population and economic statistics materials, and remote sensing image results. The update frequency is annual. Most documents contain fields such as spatial coordinates, constraint indicators, and regional scope. Units include square kilometers, hectares, population counts, and similar units. Structurally, there are both plain text descriptions and embedded tables, spatial data attachments, and charts.

Multi-source data sources require the parsing link to support attached files in different formats, including remote sensing image attachments, structured tables, and spatial coordinate text. The annual update feature means that the volume of documents parsed in a single session is large, so parsing efficiency and completeness must be balanced. Specific fields such as spatial constraint indicators and regional scope contained in documents require that contextual relevance be retained during chunking, to avoid splitting cross-page indicator descriptions. At the same time, charts and formula content in documents must retain format associations during parsing, to prevent the inability to restore data logic after chunking.

How to set configurations

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

Configuration ItemRecommended ValueRationale for This Setting
chunkSize800–1200 charactersMost documents in this category contain long-text indicator descriptions and spatial scope descriptions. This range balances contextual completeness and retrieval accuracy
chunkSplitterBy paragraph + headingDocuments in this category are layered by planning chapters and indicator items. Chunking by headings preserves chapter logic
PARSE_FILE_TIMEOUT_SECONDS600 secondsLarge planning documents take a long time to parse. This duration covers full content parsing
UPLOAD_FILE_MAX_SIZE1000 MBA single planning document may contain multiple remote sensing image attachments, so large file uploads must be allowed
customPdfParseEnabledMost documents in this category contain embedded tables and spatial coordinates. Custom parsing optimizes format restoration effects
chunkSettingModeManual configurationChunking rules need to be adjusted for different document types to adapt to multi-source data formats

Three common errors

  • Symptom: Parsing fails after uploading a docx file containing images, returning the Invalid image file error. Cause: The customPdfParse configuration is not enabled. The default parsing engine cannot support embedded remote sensing image thumbnails or spatial annotation images in planning documents.
  • Symptom: When parsing large planning documents, logs continuously report slow operation xxxxms, and only partial data is captured. Cause: The PARSE_FILE_TIMEOUT_SECONDS parameter is not adjusted. The default duration is insufficient to complete full content parsing, resulting in parsing interruption.
  • Symptom: Markdown-formatted table content is truncated when displayed after retrieval. Cause: The chunkSize configuration is too small, splitting complete table rows or cross-page indicator descriptions, resulting in loss of table context during chunking.

How to confirm configurations are set correctly

  • Upload a small planning document containing images and tables, and check whether the parsing result completely restores image placeholders and table formats.
  • Check the PARSE_FILE_TIMEOUT_SECONDS configuration item, confirm that its value is greater than 600 seconds. Verify whether parsing times out by uploading a test document larger than 100 MB.
  • Randomly select cross-page indicator descriptions from the document, and check whether the chunking result retains the front and back context of the indicators.
  • Check whether the UPLOAD_FILE_MAX_SIZE configuration is set to above 1000 MB. Try uploading a test file containing remote sensing image attachments to verify the upload and parsing process.

The values provided are common starting points and should be measured 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.