Reference only; verify against the target FastGPT release and deployment documentation.
What this category of data looks like
Coal-electric integration policy and regulatory basis data comes mainly from official public documents issued by the National Energy Administration, the State-owned Assets Supervision and Administration Commission of the State Council, and local energy regulatory authorities. It also includes industry standard documents released by the China Coal Industry Association and the China Electricity Council. Document formats include PDF, Word official documents, scanned files, and other types. The update schedule is irregular, aligned with adjustments to industry policies. Single documents typically contain structured fields such as policy document numbers, issuing authorities, effective dates, scope of application, and enforcement clauses. Some supporting interpretation documents include data tables. Units follow industry standards, such as ten thousand kilowatts for installed capacity and ten thousand tons per year for production capacity.
What constraints do these characteristics impose on document parsing and chunking
Coal-electric integration policy documents have structured characteristics and multi-format attributes. These impose multiple constraints on the document parsing and chunking process. Official policy documents have clear clause hierarchies. Chunking must retain semantic links between headings and paragraphs. Avoid splitting that breaks clause integrity. A notable share of policy documents use scanned format. The process must support OCR text extraction. Without this support, parsed content may be empty. Compilation-style policy documents have large individual file sizes. The system must adapt to longer parsing durations and higher upload limits. Structured fields such as policy document numbers must be accurately extracted. These fields serve as reference basis for Q&A.
How to set the configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | The content of single clauses in coal-electric policy documents is mostly 500-1000 characters. This range retains clause integrity |
chunkSplitter | Split by paragraph + heading | Policy documents have clear heading hierarchies and paragraph divisions. This method retains semantic coherence |
customPdfParse | Enable OCR recognition mode | Some policy files are in scanned format. OCR is required to extract text content |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Large policy compilation files take longer to parse. This setting avoids timeout interruptions |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Supports batch upload of multiple policy files, adapting to compilation-style document needs |
chunkSettingMode | Manual configuration | Flexible adjustment of chunking rules is available for the specific structure of coal-electric policies |
Three common configuration errors
- After uploading a scanned coal-electric policy PDF, the parsed result only has minor garbled text or is empty. Cause: The OCR recognition mode for
customPdfParseis not enabled. The system only supports text-based PDF parsing, and cannot extract text from scanned files. - After importing a non-standard format policy document (such as changing the file suffix from docx to txt), no valid parsed data is returned. Cause: The document format has been tampered with. Original heading and paragraph structure markers are lost. The parsing tool cannot recognize semantic division rules, so chunking cannot be completed.
- Incomplete policy basis appears during Q&A. Cause:
chunkSizeis set too small. Complete policy clauses are split into multiple independent chunks. The system cannot retrieve complete policy content as a reference basis during Q&A.
How to confirm proper configuration
- Upload a standard coal-electric policy PDF file. Check if the parsed text includes complete structured fields such as policy document numbers, issuing authorities, and effective dates.
- Adjust the
chunkSizeparameter to different value ranges. Compare the chunked results to confirm that single chunks do not break policy clause integrity. - Upload a scanned coal-electric policy PDF. Verify that complete text content is extracted normally in OCR mode.
- Upload a large policy compilation file. Confirm the parsing process does not time out, and chunked results meet expectations.
The values provided are common starting points and should be evaluated 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.