Reference only; verify against the target FastGPT release and deployment documentation.
What This Category of Data Looks Like
This category of data originates from the China Disease Prevention and Control Information System, medical institution HIS and infectious disease reporting systems, and laboratory testing systems. The update frequency is daily. Commonly processed documents include standardized infectious disease reporting cards, epidemic analysis reports, case correction records, clustered epidemic reports, and epidemiological investigation reports. Document structures include core business fields such as basic case information, diagnosis results, reporting time limits, correction notes, and laboratory test data, with no extra non-business redundant content.
Constraints Imposed on Document Parsing and Chunking by These Characteristics
Daily batch-updated documents require parsing processes with stable batch processing capabilities, to avoid delaying epidemic monitoring timeliness. Standardized reporting cards and documents with correction records coexist, requiring parsing links to accurately identify field and version differences to avoid missing correction information. Documents from multiple systems have slight format differences, requiring parsing logic adaptable to different input formats. Since data is linked to legal infectious disease reporting rules and epidemic early warning requirements, parsing and chunking links must retain contextual associations of key fields, and core business information cannot be truncated arbitrarily.
How to Set Configurations
| Config Item | Recommended Value | Basis for This Value |
|---|---|---|
chunkSize | 800–1200 characters | Adapt to single-field or short-paragraph business units of infectious disease reporting cards and epidemiological investigation reports, avoid truncating key information such as case diagnosis and reporting time limits |
chunkSplitter | Split by paragraph + semantic boundary | Adapt to format differences of documents from multiple systems, retain complete semantics of correction records and clustered epidemic descriptions |
customPdfParse | Enabled | Adapt to scanned PDF documents (such as epidemiological investigation records uploaded as scanned paper reports), extract text content from images |
PARSE_FILE_TIMEOUT_SECONDS | 120–180 seconds | Respond to large-volume batch epidemic analysis reports, avoid task failure due to parsing timeout |
chunkSettingMode | Manual configuration | For the particularity of fields in legally required reporting documents, flexibly adjust chunking rules to match business verification needs |
Three Common Mistakes
- Uploading a scanned PDF document results in empty parsing output or only a small amount of extracted text. This occurs when the
customPdfParseconfiguration is not enabled, so the OCR text extraction process is not triggered. - Timeout errors appear when parsing batch infectious disease reporting cards. The error message shows request timeout. This occurs when the
PARSE_FILE_TIMEOUT_SECONDSsetting value is lower than the actual required parsing duration. - Case diagnosis information is truncated in post-chunking search results. This occurs when the
chunkSizevalue is too small, truncating the complete business unit containing diagnosis results and reporting time limits.
How to Confirm That Configurations Are Correct
- Upload a single scanned epidemiological investigation report PDF, check if the parsing result includes core fields such as basic case information and diagnosis results, confirm that the configuration takes effect.
- Upload batch infectious disease reporting card documents, check the time consumption records of parsing tasks, confirm that the timeout configuration setting value is higher than the actual average parsing duration.
- Extract the parsing result of a long document, check if the chunked content retains complete case correction notes or clustered epidemic descriptions, confirm that the chunking rules adapt to business needs.
- Test Feishu multidimensional table data source access, complete the authorization and docking configuration required by the platform, verify that data reading is normal.
All provided configuration values 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.