What the data for this category looks like
Data for this category mainly comes from the national unified construction project EIA declaration and approval system, ecological environment zoning control results, pollution discharge permit platform, and monitoring data. The update frequency is irregular. Daily processed documents include environmental impact assessment reports/report forms, approval applications, public participation statements, technical evaluation opinions, EIA approvals, and similar materials. Their structures usually contain modules such as basic project information, pollution prevention analysis, public participation status, approval process opinions, and others. Some documents include non-standardly formatted tables and attached figure descriptions.
What constraints do these characteristics impose on the "document parsing and chunking" link
The multi-type, long-form, and non-standard formatting characteristics of EIA documents create multiple constraints for document parsing and chunking. First, a single EIA report may contain dozens of pages. It is necessary to balance chunk length and context integrity. Avoid splitting chunks too finely to lose business connections, or making chunks too long to reduce subsequent retrieval accuracy. Second, document formats from multiple sources are inconsistent. Some documents contain embedded tables, handwritten annotations, or text converted from scanned copies. Text extraction for non-standard formatting must be supported. Third, the business link requires accurate extraction of core content such as project category and pollution prevention measures. Chunking must retain the context of corresponding modules to ensure continuity of subsequent verification and approval work. In addition, parsing timeout or errors directly affect the progress of the approval process. Parsing stability must be guaranteed.
How to set the configurations
| Configuration Item | Recommended Approach | Basis for This Approach |
|---|---|---|
chunkSize | 800–1200 characters | EIA documents contain long technical analysis paragraphs. This length preserves context integrity while avoiding excessively large chunks that reduce retrieval efficiency |
chunkSplitter | Split by Markdown headings + paragraphs | EIA documents have clear chapter heading divisions. Splitting by headings preserves the integrity of business modules and adapts to different document structures |
customPdfParse | Enabled | EIA PDFs often contain non-standardly formatted tables, attached figure descriptions, and hidden text. Custom parsing can fully extract all valid content |
PARSE_FILE_TIMEOUT_SECONDS | 1200 seconds | Large EIA reports take a long time to parse. This timeout setting prevents task failure due to insufficient parsing duration |
UPLOAD_FILE_MAX_SIZE | 100 MB | Single EIA reports usually do not exceed this size, adapting to document upload requirements for most business scenarios |
chunkSettingMode | Manual configuration | EIA document formats vary widely. Manual configuration can accurately match the chunking needs of different document types |
Reference only; verify against the target FastGPT release and deployment documentation.
Three common mistakes
- Phenomenon: A timeout error is prompted when uploading an EIA PDF larger than 10 MB. Cause: The
PARSE_FILE_TIMEOUT_SECONDSparameter is not adjusted. The default timeout duration is insufficient to parse large documents. - Phenomenon: Extremely long technical paragraphs appear after chunking and are not split again. Cause: The
chunkSplitteris not configured correctly. Only fixed delimiters are used for splitting, and extremely long paragraphs are not processed. - Phenomenon: Text from the public participation statement module is missing in the parsing result. Cause:
customPdfParseis not enabled. It is impossible to extract attached text with non-standard formatting.
How to confirm the configuration is correctly set
- Upload a single typical EIA document, view the parsed chunk list, confirm that each chunk contains complete business module content, and adjust
chunkSizeuntil the chunk length meets expectations. - Upload an EIA PDF larger than 10 MB, check whether the parsing task is completed within the preset timeout duration, and verify the effectiveness of the
PARSE_FILE_TIMEOUT_SECONDSconfiguration. - Import a custom parsing script, test whether the parsing result contains hidden table note text from the original document, and confirm that the
customPdfParseconfiguration takes effect. - Test multiple EIA documents of different formats (such as report forms and full reports), confirm that the chunking logic adapts to different document structures, and adjust the
chunkSplitterparameter to match business needs.
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.