What the data for this category looks like
Core documents for this category include property right registration forms, asset appraisal reports, property right transfer plans, on-site transaction announcements, asset disposal approval forms, and similar materials. Data primarily comes from state-owned asset supervision information systems, enterprise asset management systems, property right exchange systems, appraisal agency reports, and financial statements. The data updates monthly. Most documents mix structured fields and unstructured text. Some asset appraisal reports include multi-page tables and nested annotations, with fields covering exclusive attributes such as asset number, ownership unit, and appraisal value.
What constraints do these characteristics impose on the document parsing and chunking process
The mixed structure, exclusive field attributes, and multi-source nature of documents in this category create multiple constraints for the parsing and chunking workflow. Documents from multiple systems have inconsistent formats, so the parsing process must support requests for multiple file types including PDF, Word, and Excel. Documents that mix structured fields and unstructured text require the parsing workflow to retain table structures and field associations, and avoid splitting key information such as asset numbers and ownership units. The monthly batch update requirement means the parsing link needs stable timeout control to prevent batch task blocking. For asset appraisal reports with multi-page nested annotations, precise page boundary identification is needed to avoid splitting a single appraisal entry across pages.
How to set configurations
The following configurations apply to FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Documents in this category include multi-page tables and exclusive fields. This range retains the full context of information tied to a single asset, preventing key content from being split |
chunkSplitter | Separate by paragraph + title | Documents in this category use a business structure with hierarchical titles. Splitting by title preserves the logical connection of business workflows such as property right transfer and asset disposal |
customPdfParse | Enabled | Asset appraisal reports in this category mostly include nested annotations and complex tables. Enabling the custom PDF parsing service retains original typesetting and table structure, and supports local deployment to avoid file content leakage |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | A single asset appraisal report in this category may include multi-page complex content, requiring sufficient time to complete full parsing |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Some batch-uploaded state-owned asset supervision report collections in this category have a large total size, requiring support for large-file uploads |
Three common misconfigurations
- Symptom: Tables are parsed incorrectly after uploading a PDF file, and the asset appraisal value field in the table is missing from chunking results. Cause: The
customPdfParseconfiguration is not enabled. The default general parsing engine cannot adapt to the complex table structure of documents in this category. - Symptom: File parsing fails when uploading a file during a conversation, returning a timeout error or status code 504. Cause: The
PARSE_FILE_TIMEOUT_SECONDSparameter is not adjusted. The default duration is insufficient for parsing multi-page asset appraisal reports in this category. - Symptom: Chunking results concatenate content across assets, breaking field associations. Cause: The
chunkSizevalue is too small, splitting asset numbers and transfer conditions in the same property right transfer plan into different chunks.
How to confirm configurations are correct
- Upload a typical asset appraisal report for this category, verify that the parsed text retains complete table structure and annotation content, and confirm the
customPdfParseconfiguration is active. - Upload a single large-sized state-owned asset supervision report collection, check that upload progress and parsing results function normally, and confirm the
UPLOAD_FILE_MAX_SIZEconfiguration meets requirements. - Randomly select a property right transfer plan, review chunked content to confirm it retains the connection between titles and corresponding paragraphs, and verify the
chunkSizeandchunkSplitterconfigurations are reasonable. - Submit a batch parsing task, monitor task execution duration, and confirm the
PARSE_FILE_TIMEOUT_SECONDSparameter covers the longest required parsing time.
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.