What data in this category looks like
Documents in this category mainly come from state-owned assets supervision information systems, enterprise asset management systems, property rights exchange systems, evaluation agency reports and financial statements. The update frequency is monthly. Daily processed documents include property rights registration forms, asset evaluation reports, property rights transfer plans, on-site transaction announcements, and asset disposal approval forms. Most contain structured fields such as asset numbers, ownership certificate numbers, evaluation values, and transaction base prices. Some documents include tables and long text passages. Units involved include ten thousand yuan, square meters, pieces and other measurement standards.
What constraints these characteristics impose on the document parsing and chunking link
Monthly updated batch documents require parsing processes to have stability and batch processing capability. Multi-type document structures require chunking logic to balance the integrity of structured tables and long text passages. It must avoid splitting that breaks the corresponding relationship between key fields such as asset numbers and evaluation values. Large-volume asset evaluation reports and transfer plans will increase parsing time. Longer timeout thresholds must be adapted. Excel-format asset lists need to be split by row. Otherwise, single records may span multiple chunks, which affects subsequent retrieval accuracy.
How to set the configuration
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value Range | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the multi-field, long-paragraph structure of documents in this category, avoiding splitting that breaks the integrity of key fields such as asset numbers and evaluation values |
chunkSplitter | Mixed splitting by tables and passages | Matches content that contains both text passages and structured tables in asset evaluation reports and property rights transfer plans, avoiding splitting table rows into multiple chunks |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Addresses parsing time for large files such as single 10–20 MB asset disposal approval forms and property rights transaction announcements |
UPLOAD_FILE_MAX_SIZE | 20 MB | Adapts to the common single-document volume upper limit for this category, avoiding large file upload failures |
customPdfParse | Enabled | Retains table structures and text hierarchy in PDF versions of asset evaluation reports, avoiding chaotic content after parsing |
Three common mistakes
- Symptom: Excel-imported asset lists are merged into a single chunk, with single asset records scattered across multiple chunks. Cause:
chunkSplitteris not configured to split structured tables by row, and the default chunking logic does not recognize Excel row boundaries. - Symptom: Large-volume Word-format asset evaluation reports fail parsing due to timeout. Cause:
PARSE_FILE_TIMEOUT_SECONDSis not adjusted to a value suitable for large files, and the default timeout duration is insufficient. - Symptom: Table fields are missing after parsing PDF versions of property rights registration forms. Cause:
customPdfParseis not enabled, and the default parsing logic loses the column and row corresponding relationships of structured tables.
How to confirm the configuration is correct
The configuration effect can be verified through the following actions in FastGPT v4.17.0:
- Upload a single asset evaluation report of approximately 15 MB, and check whether parsing completes within 600 seconds with no timeout error prompts.
- Upload an Excel asset list containing structured tables, and check whether each chunk corresponds to one independent asset record in the chunking results.
- Upload a PDF version of a property rights transfer plan, and confirm that the parsed content retains the original row and column structure of the table with no text misalignment or field loss.
- Check whether the uploaded file size limit is set to 20 MB, and attempt to upload a 20 MB document with no upload failure prompts.
The values provided are common starting points and should be measured against samples relevant to the specific use case.
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.