What the data for this category looks like
The business data for this specific scenario mainly comes from national public resource trading platforms, local public resource trading platforms, engineering approval systems, and bidding and tendering supervision systems. It is updated daily. Commonly processed documents include bidding announcements, qualification prequalification documents, tender documents, bid documents, bid evaluation reports, successful bid candidate notices, and successful bid result announcements. Most document structures include fields such as section division, qualification requirements, bill of quantities, and bid evaluation rules. Some documents contain multi-column tables, nested paragraphs, and structured data.
What constraints these characteristics impose on the "document parsing and chunking" link
The daily updated batch of documents requires the parsing link to support efficient batch processing, to avoid delays in public resource trading business processes. The presence of multiple types of structured documents and multi-column tables requires the chunking logic to adapt to table rows, paragraph boundaries, and business unit divisions such as sections and qualifications, to avoid splitting that breaks business relevance. Bidding and tendering documents have high compliance requirements. Chunking results must retain complete key information such as qualification conditions and bid evaluation standards, to prevent compliance risks caused by information loss from splitting. Additionally, document lengths vary significantly, from short successful bid announcements to tens of thousands of words in bidding documents. Flexible adjustment of chunking parameters is needed to cover all types of bidding documents.
How to set the configuration
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the business unit length of long documents such as bidding documents and tender documents, avoiding splitting that breaks the integrity of key information such as sections and qualifications |
chunkSplitter | Split by paragraphs and table rows | Adapts to the characteristics of multi-column tables and structured paragraphs in bidding documents, meeting the need for precise segmentation by business units |
customPdfParse | Enabled | For formatted PDF documents such as bidding announcements and bid evaluation reports, retains original typesetting and table structures to improve parsing accuracy |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Addresses the parsing time requirements of tens of thousands of words in bidding documents, avoiding parsing failures caused by timeouts |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapts to scenarios of batch uploading large bidding documents and tender documents, meeting the needs of full business data import |
chunkSettingMode | Manual configuration | Supports adjusting chunking parameters for different document types, adapting to the parsing needs of multiple types of bidding documents |
Three Common Mistakes
- Phenomenon: Auto-chunking results for multi-column bidding Excel files (such as bill of quantities) are chaotic, and cannot split by single rows. Cause: The default chunking logic does not adapt to row-level splitting rules for multi-column tables, and only splits by general paragraph boundaries.
- Phenomenon: When parsing a single uploaded attachment, the system loads previously uploaded bidding documents at the same time, causing chunking results to include irrelevant data. Cause: Single-document independent parsing mode is not enabled, and the historical parsing cache of the global knowledge base is reused by default.
- Phenomenon: Images in bid evaluation reports and qualification prequalification documents in bidding documents cannot be parsed, and key information in images cannot be extracted. Cause: Image parsing configuration is not enabled, or the uploaded format does not meet supported image type standards.
How to Confirm the Configuration is Correct
- Upload a standard bidding document PDF, review the parsed chunking results, and confirm that key business units such as section division and qualification requirements are not split or damaged.
- Upload a single Excel bill of quantities, verify that chunks are generated per single row, with no chaotic segmentation across rows or columns.
- Upload a bid evaluation report with images, confirm that the parsing result includes OCR text content of the images, with no missing content.
- Check the parsing task logs, and confirm that no timeout errors exceeding the
PARSE_FILE_TIMEOUT_SECONDSthreshold occur.
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.