What Data for This Category Looks Like
Documents for policy and regulatory basis of coal transportation and sales mainly come from national energy authorities, provincial coal industry regulatory departments, and enterprise internal compliance departments. Most documents follow official document formats, with fields such as release document number, effective date, applicable region, and specific transportation and sales clauses. Some attachments include data such as transportation volume calculation tables and freight rate standards. National-level policy updates do not follow a fixed cycle. Enterprise internal systems are revised quarterly or annually. The length of individual documents ranges from 5000 to 20000 characters. Some long documents include multi-level nested chapters.
Constraints Imposed by These Characteristics on Document Parsing and Chunking
The multi-level nested chapter structure of coal transportation and sales policy documents requires retaining context associations during chunking. Avoid splitting a single compliance clause. The length of individual documents varies widely. Using fixed character chunking for overly long documents may split associated data such as transportation volume and freight rates. Metadata for policy documents, such as document number and effective date, must be bound to the main text. Otherwise, the applicable scope of the corresponding clause cannot be accurately associated. For attachment tables included in some documents, the corresponding relationship between cells must be retained during parsing. Otherwise, subsequent Q&A cannot accurately match data. The frequent revision of enterprise internal systems also requires the parsing process to quickly adapt to changes in document formats.
Configuration Settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Single clauses in coal transportation and sales policies range from 300 to 800 characters. This range can cover complete clauses while avoiding overly long chunks |
chunkSplitter | Mixed chapter and fixed character chunking | Documents have multi-level nested chapters. The mixed mode can retain structure while avoiding overly large single chunks |
customPdfParse | Enabled | Most coal transportation and sales policy documents include official document numbers and attachment tables. Custom parsing can accurately extract metadata and table content |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Some annual coal transportation and sales policy compilation documents are lengthy. Large file upload support is required |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | The chapter recognition and chunking process for long documents requires sufficient time to complete processing |
chunkSettingMode | Manual configuration | Chunking rules must be adjusted for the chapter structure and field characteristics of coal transportation and sales documents |
Three Common Mistakes
- Scenario: After uploading a Word-format coal transportation and sales policy document, no image links are extracted in the parsing results. Image content cannot be called during Q&A testing. Cause: The image extraction configuration for
customPdfParseis not enabled. Only text content is extracted. - Scenario: Parsing errors occur due to timeout when processing large annual policy compilation documents. Cause: The value set for
PARSE_FILE_TIMEOUT_SECONDSis lower than the actual time required for document parsing. - Scenario: A single coal transportation and sales compliance clause is split into two separate chunks in the chunking results. Cause:
chunkSplitteronly uses fixed character chunking, and does not adapt to the chapter structure of the document.
How to Verify Proper Configuration
- Upload a typical coal transportation and sales policy official document. Check whether metadata such as document number and effective date is fully extracted in the parsing results.
- Test the chunking results. Confirm that a single compliance clause is not split into multiple chunks.
- Upload a document with an attachment table. Confirm that table content is fully retained and associated with the corresponding parsing chunk.
- Upload a document with a length exceeding 10000 characters. Confirm that the parsing process does not trigger a timeout error.
The values provided here are common starting points. They 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.