What data for this category looks like
Data for this category primarily comes from switch consoles, wireless controllers, office area monitoring systems, and end-user trouble reporting channels. Daily produced documents include network topology diagrams, network change orders, troubleshooting records, wireless coverage inspection forms, and more. Data is updated daily or immediately following network changes. Most document fields include port numbers, VLAN identifiers, signal strength values, inspection points, fault location information, and other details. Some documents are in exported binary formats or have non-standardized layouts.
What constraints these characteristics place on the document parsing and chunking step
Multi-source heterogeneous document formats require the parsing module to adapt to different file types, and avoid losing key information such as vector annotations in network topology diagrams and table fields from Excel inspection forms. High-frequency updated real-time data requires controlling parsing latency to prevent document upload failures due to timeouts. For documents containing professional configuration fields, chunking must retain contextual associations, to avoid breaking the logical connections of configuration items such as ports and VLANs after splitting. Scattered user trouble report texts require precise extraction of paragraphs related to fault scenarios, to prevent irrelevant information from mixing into chunks and affecting subsequent question-and-answer accuracy.
How to set configurations
All following configurations are based on FastGPT v4.17.0.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
chunkSize | 800–1200 characters | Adapts to the paragraph length containing multiple fields in network documents, retains contextual associations of configuration items such as ports and VLANs |
chunkSplitter | Split by paragraph + heading | Matches the layout logic of structured documents such as network topology diagram descriptions and change order chapters |
customPdfParse | Enable vector text extraction mode | Retains key information such as coordinates and port annotations in network topology diagrams, avoids losing graphic association data during plain text parsing |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Addresses parsing time requirements for large network topology PDF files or batch inspection form Excel files |
UPLOAD_FILE_MAX_SIZE | 1000 MB | Compatible with upload requirements for single large network topology collections or quarterly inspection summary forms |
chunkSettingMode | Manual configuration mode | Adjusts chunking rules separately for different types of operation and maintenance documents, adapts to heterogeneous document formats |
Three common mistakes
- After uploading a PDF network topology diagram, the knowledge base cannot recognize port annotation information. The cause is that the vector text extraction mode of
customPdfParseis not enabled, and only plain text content is extracted, resulting in lost graphic association fields. - After configuring batch upload of inspection form Excel files, the generated SQL cannot be queried normally. The cause is that
chunkSettingModeis not adjusted to adapt to table fields, and complete configuration items in cells are split during chunking, leading to broken field associations. - After uploading a single fault record document, the question-and-answer return result only contains scattered sentences. The cause is that
chunkSizeis set too small, splitting the complete fault troubleshooting process into multiple unrelated short chunks, losing contextual logic.
How to confirm configurations are applied correctly
- Upload a typical network topology diagram PDF, check if the parsed text includes annotation information such as port numbers and VLAN identifiers, to confirm that the
customPdfParseconfiguration is effective. - Upload a batch inspection form Excel file, verify that the chunked content retains the complete field combination of a single inspection form, to confirm that the
chunkSplitterrule matches the document structure. - Submit a query question related to network configuration, check if the returned result includes a complete fault troubleshooting process or topology association information, to confirm that the
chunkSizesetting is reasonable. - Upload a single large change order document, check if the parsing progress is completed within the time set by
PARSE_FILE_TIMEOUT_SECONDS, to confirm that the timeout configuration adapts to the document size.
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.