What the data for this category looks like
Steel trade intelligent due diligence reports are business documents used in financial supply chain scenarios. Data sources primarily include internal inventory ledgers of traders, factory delivery documents from steel mills, port pickup vouchers, downstream purchase orders, and settlement vouchers. Data is updated on a daily basis. Core fields include steel grade, nominal diameter, theoretical weight, tax-included unit price, pickup date, ownership voucher number, and payment voucher number. Units include ton, yuan/ton, yuan, and others. Common document formats are multi-page structured Excel or scanned PDF files; some ownership vouchers with handwritten annotations count as semi-structured data.
What constraints these characteristics impose on deployment and upgrade
The mixed multi-source semi-structured and structured data formats require adapting to multiple types of file parsing rules during deployment. When upgrading, existing parsing plugins must be retained to maintain compatibility with historical ledger formats. Daily high-frequency updated business data requires vector database caching strategies adapted for high-volume writes. During upgrades, cache configuration resets that cause data synchronization interruptions must be avoided. Fields with fixed names but mixed units require configuring field mapping rules during deployment. When upgrading, field matching logic must not be overwritten. Bulk large-file upload requirements necessitate relaxing upload and parsing timeout limits. During upgrades, related parameters must be adjusted synchronously to ensure business continuity.
How to set configurations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Steel trade due diligence reports often include multi-page inventory ledgers and ownership vouchers, leading to long single-file parsing times |
UPLOAD_FILE_MAX_SIZE | 2000 MB | Bulk uploaded trade contracts and ledger files have large individual file sizes, requiring relaxed upload limits |
similarityThreshold | 0.72–0.80 | Approximate wording exists for steel grade and specification fields, requiring threshold adjustments to ensure accurate matching |
recallTopK | Top 8 entries | Due diligence reports need to cover multi-dimensional upstream and downstream data, requiring sufficient associated documents to be recalled |
MCP_SERVER_ENABLED | false | Basic due diligence scenarios do not require calling external tools, so this can be disabled to reduce deployment dependencies |
REDIS_MAX_MEMORY | 10 GB | Stores session and cached data, adapting to the caching needs of high-frequency updated trade data |
The parameter values provided on this page are common recommended starting points for configuration. Actual values are affected by material format, data volume, and business rules. Specific issues require individual analysis, and it is recommended to test on your own samples before finalizing settings.
Three common configuration mistakes
- After upgrading to version 4.9.2, starting the service prompts that the
systemEnv.pconfiguration item is undefined, and no corresponding field exists inconfig.json. The cause is that this version migrated some system environment variables from thesystemEnvhierarchy inconfig.jsonto root-level environment configuration files. - When accessing a due diligence report via a non-login share link, the "View Original" button is grayed out and unavailable. The cause is that version 4.9.6 disables public share document viewing permissions by default, requiring manual adjustment of the corresponding configuration item.
- After bulk uploading multiple trade ledgers, all parsing tasks time out and fail. The cause is that the
PARSE_FILE_TIMEOUT_SECONDSparameter was not adjusted; the default timeout duration is insufficient for parsing large multi-page files.
How to confirm configurations are correctly set
- Upload a standard steel trade ledger file, check if the parsed fields include the preset business fields, and verify that the field mapping matches actual business requirements.
- After starting the service, check the runtime logs to confirm that the Redis connection is normal, with no out-of-memory or connection timeout errors.
- Test access permissions via the non-login share link, confirm that the "View Original" function triggers normally and displays the original document content.
- Initiate a bulk parsing task, confirm that the task completes within the preset timeout duration, with no bulk task backlog.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.