Reference only; verify against the target FastGPT release and deployment documentation.
What the data for this category looks like
The core data for this category comes from national petition information systems, local petition systems, the 12345 citizen hotline platform, grassroots governance platforms, and historical petition archives. Data updates occur daily. Daily processed documents include standard formatted files such as petition registration forms, acceptance notices, transfer and assignment letters, handling opinions, review and re-examination applications, and resolution plans. Some files are historical archives stored as offline scanned documents. Data fields include basic petitioner information, appeal content, handling nodes, responsible units, handling opinions, and other items. There is no unified fixed unit field. Only business-related information is associated based on the active business scenario.
Constraints on the document parsing and chunking link
Daily updated batch data requires the parsing link to support efficient batch processing. This prevents delays that disrupt ledger summary progress. Mixed sources of standard formatted documents and scanned archives require the parsing module to support multiple input formats. The module must also retain legally required handling step information from documents. This avoids errors in handling procedures caused by parsing omissions. The lack of a unified fixed unit field means chunking must not forcibly bind unit information. Only original business content should be retained to preserve data integrity. Errors in petition data directly relate to issues such as repeated petitions and petitions submitted beyond the proper hierarchy. Therefore, chunking must ensure that key node information is not split or lost.
How to configure the settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Petition-related documents contain both long-text appeals and short-field handling nodes. This range covers core information and avoids splitting key content |
chunkSplitter | Calibrated based on actual testing | Petition documents from different sources have large format differences. Splitting by headings and line breaks retains business logic integrity |
customPdfParse | Enabled | Most petition-related PDFs are standard table-based documents. Enabling this allows accurate extraction of fields such as handling document numbers and responsible units within tables |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Batch-imported petition archive PDFs may contain multiple pages, requiring sufficient time to complete full parsing |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Adapts to the upload requirements of single large-scale petition archive scanned files |
chunkSettingMode | Manual configuration | For petition documents from different sources, manually adjusting parameters achieves more accurate chunking effects |
Three common configuration mistakes
- Uploading non-standard petition documents with modified file extensions, such as changing files of other formats to .txt and importing them. The parsing result will have no valid content. Reason: Custom parsing configuration is not enabled. Only the default suffix matching logic is used, which cannot recognize non-standard text formats.
- Calling the pushdata API with the
chunkmode to upload data, and the interface displays the "indexing" status for a long time. Reason: A reasonablePARSE_FILE_TIMEOUT_SECONDSparameter is not set, or the number of documents uploaded in a batch exceeds the single-processing limit. - Uploading a petition PDF containing scanned content. The parsing result does not extract text information from the image. Reason: OCR parsing configuration is not enabled. The default PDF parsing only extracts text layer content, and cannot recognize text within scanned images.
How to confirm the configuration is correct
- Upload a standard petition registration form PDF. Check whether core fields such as petitioner name, appeal content, and handling unit are extracted after parsing.
- Call the pushdata API to upload a single test document. Check whether the parsing time falls within the range set by
PARSE_FILE_TIMEOUT_SECONDS. - Adjust the
chunkSizeparameter. Compare whether the chunking result retains complete handling step information, with no key content split. - After enabling
customPdfParse, upload a standard petition PDF. Confirm that fields such as handling document numbers and responsible units within the table are correctly extracted.
The values provided here are common starting points, and should be evaluated against the user'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.