What data looks like for this category
Daily documents for this category include four core types: procurement requests, price comparison records, contracts, and acceptance documents. Data sources cover internal procurement systems, supplier quotation documents, budget spreadsheets, and contract systems. Data updates follow a per-order/per-month rhythm: each individual procurement process corresponds to a separate set of documents. Most document structures include both structured fields and long-form text descriptions. Procurement requests contain requirement specifications, budget amounts, and alternate supplier lists. Contracts include compliance clauses, delivery milestones, and breach clauses. Price comparison records contain quotation details and qualification attachments, with units and identifiers such as RMB amounts, calendar days, and qualification numbers.
What constraints these characteristics impose on the document parsing and chunking link
Heterogeneous document formats from multiple sources create parsing adaptation pressure. These include structured tables exported from procurement systems, PDF contracts, and Word quotation documents sent by suppliers, so parsing rules must be adapted for different formats. The volume of documents in a single batch fluctuates with procurement orders, ranging from 1 to dozens of documents per single processing. Chunking logic must balance single-block accuracy and batch processing efficiency. Compliance-related clauses must be extracted accurately, with no irrelevant content included in chunks. Otherwise, retrieval matching for subsequent compliance reviews will be affected. Data updates are triggered per batch, so no full incremental sync is required. Only documents for the current procurement order need parsing and chunking.
How to set configurations
These configurations are based on the functionality logic of FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Procurement documents mostly include structured clauses and long text descriptions. This range preserves clause integrity while avoiding overly long single blocks that hinder retrieval |
chunkSplitter | Split by paragraph + heading | Procurement documents have clear chapter divisions, such as procurement requirements and compliance clauses. Splitting by headings preserves the logical structure of the document |
PARSE_FILE_TIMEOUT_SECONDS | 120 seconds | Parsing takes longer when a single procurement order’s documents include multiple attachments. This duration prevents parsing timeout interruptions for standard procurement documents |
UPLOAD_FILE_MAX_SIZE | 500 MB | Accommodates the file size of bulk uploaded supplier quotation collections and large contract PDFs, meeting file upload requirements for procurement scenarios |
customPdfParse | Enabled | Procurement PDF contracts often have fixed formats. Custom parsing can accurately extract compliance clause fields and filter irrelevant content such as headers and footers |
chunkSettingMode | Manual configuration | Document formats vary significantly across different procurement orders. Manual configuration adapts to the document characteristics of specific procurement scenarios |
Three common mistakes
- Scenario: A large procurement contract PDF is blocked after upload, with an on-screen prompt that the file exceeds size limits. Cause: The
UPLOAD_FILE_MAX_SIZEparameter was not adjusted, and the default value cannot accommodate large-volume documents in procurement scenarios. - Scenario: Consecutive compliance clauses in the parsed chunk results are split across multiple blocks, preventing complete matching during retrieval. Cause:
chunkSizeis set too small, not matching the clause length of procurement contracts, leading to forced splitting of compliance content. - Scenario: Chunk index fields are missing from parsing results retrieved via API calls, with no corresponding identifiers in returned content. Cause: Chunk index output configuration was not enabled, and the indexing function for the corresponding parameters was not activated as required by the API.
How to confirm configurations are set correctly
- Upload a standard procurement contract PDF, check if parsed chunks are split by chapter, and whether compliance clauses are fully retained within a single block.
- Upload a batch of procurement request spreadsheets, verify that no file upload blocking prompts appear, and confirm that the
UPLOAD_FILE_MAX_SIZEconfiguration takes effect. - View parsing logs to confirm there are no timeout errors, verifying the effect of the
PARSE_FILE_TIMEOUT_SECONDSparameter configuration. - Call the API to retrieve chunk results, check if chunk index fields are included, and confirm that the configuration meets API call requirements.
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 2026-10-19.