What this category's data looks like
Data for this category primarily comes from ERP material master data, melting furnace thermometers, spectrometers, MES heat records, mold management systems, and customer technical drawings. Daily documents include foundry process cards, batching lists, heat records, mold histories, first article inspection records, material test reports, and nonconforming product review forms. Update frequency is per heat, per batch, or daily. Most documents contain fields such as alloy ratio, pouring temperature, mold number, heat number, and defect type. Units include degrees Celsius, percentage, kilograms, and others. Some process cards have non-standard table layouts.
What constraints do these characteristics impose on the document parsing and chunking link
The high-frequency update and multi-field associated document characteristics bring multiple constraints to document parsing and chunking. First, documents such as heat records and batching lists are organized by heat or mold number. When splitting chunks, retain the integrity of single-heat data, and avoid mixing cross-batch content. Second, large historical summary documents may exceed default parsing time limits, so adjust timeout parameters. Third, non-standard process card tables cannot extract fields accurately via general parsing, so enable custom parsing adaptation. Bulk uploads of multiple documents need to support large queues to prevent upload failures.
Configuration settings
Configuration settings for FastGPT v4.17.0 are as follows:
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Matches the length of single process parameters or defect records in foundry documents, avoids splitting cross-heat content |
chunkSplitter | Custom split by heat number, mold number | Foundry documents are mostly organized by heat/mold number; using dedicated identifiers ensures data relevance within chunks |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Handles parsing of heat summary records over 10MB, prevents timeout errors |
customPdfParse | Enabled | Adapts to non-standard table layouts in foundry process cards, preserves original field formatting |
UPLOAD_FILE_MAX_SIZE | 2000 MB | Supports bulk uploads of historical records across multiple heats |
Three common configuration errors
- Uploading a heat summary PDF over 10MB triggers the prompt
timeout of 360000ms exceeded. The cause is failure to adjust thePARSE_FILE_TIMEOUT_SECONDSparameter. The default timeout duration is insufficient for parsing large foundry documents. - Parsed chunks contain cross-heat batching and inspection data. The cause is failure to use heat number as the
chunkSplitter. Default general punctuation-based splitting causes mixed data within chunks. - Fields such as alloy composition and pouring temperature are empty after parsing. The cause is failure to enable
customPdfParse. Default parsing cannot recognize embedded table fields in foundry process cards.
How to confirm correct configuration
- Upload a single 10MB heat record PDF, and check if the parsing task completes within 10 minutes with no timeout errors.
- Randomly select 3 documents from different heats, and check if parsed chunks only contain single-heat batching, temperature, and defect data.
- Open the custom parsing settings page, confirm that
customPdfParseis enabled andchunkSplitteris configured with dedicated identifiers such as heat number and mold number. - Upload bulk historical documents, and check if the upload queue supports file uploads within 2000 MB per batch.
The values listed above are common starting points. They should be validated against applicable samples.