What this category of data looks like
This category of data covers documents related to the full lifecycle of special equipment, including use registration certificates, periodic inspection reports, maintenance records, hidden danger rectification notices, accident reports, and operator certificates. Data sources include the special equipment safety supervision system, inspection and testing institution systems, user unit ledgers, and operator certificate databases. The update frequency is monthly. Each individual document contains core fields such as equipment number, using unit, inspection cycle, maintenance date, and qualification validity period. Different document types have different field focuses. For example, operator certificates prioritize certificate number and validity period, while inspection reports prioritize the next inspection date.
What constraints do these characteristics impose on the knowledge base retrieval and recall link?
Multiple data sources require retrieval to associate multiple independent datasets, to avoid missing cross-system device-related information. The monthly update frequency requires the knowledge base synchronization cycle to match the data update rhythm, ensuring recalled ledger and report information is up to date. Diverse document types and differentiated fields require configuring dedicated retrieval rules for different documents. For example, prioritize matching the validity period field for operator certificates, and prioritize matching the next inspection date field for inspection reports. Ledger summary requires batch association of multiple document types such as equipment, inspection, maintenance, and rectification. This requires that recall results can be grouped by device identifiers to facilitate subsequent summary statistics.
How to configure the settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the field precision of special equipment supervision documents, avoiding low-relevance non-supervision documents from being included |
limit | Top 10 entries | Ledger summary requires batch association of multiple types of documents such as inspection, maintenance, and rectification for a single device. A single recall is sufficient to cover core supervision records |
searchMode | hybrid | Combines keyword and semantic retrieval, adapting to structured fields such as device number and inspection date, as well as unstructured content such as maintenance descriptions |
usingReRank | Enabled | Rearranges recall results by business relevance, prioritizing results that match core supervision fields such as inspection cycle and qualification validity period |
datasetSearchUsingExtensionQuery | Enabled | Expands the retrieval scope to cover documents in different formats imported from multiple source systems, avoiding missing cross-system associated data |
collectionIds | List of dataset IDs configured for the supervision domain | Accurately limits the retrieval scope to exclusive datasets for special equipment registration, inspection, maintenance, and operator certificates, excluding irrelevant data |
Three common misconfigurations
- Symptom: An error of
fail to create post presigned urlorfailed to create post pis displayed when uploading a text dataset. Cause: The access key and bucket permissions of the S3 storage are not configured correctly, or the dataset storage path is configured incorrectly, resulting in the inability to generate a pre-signed upload link. - Symptom: Retrieval results cannot be associated with the inspection and maintenance records of the corresponding device. Cause: The retrieval scope is not limited to special equipment-specific datasets via
collectionIds, or thesimilaritythreshold is set too low, leading to irrelevant results being mixed in and failing to accurately match device identifier fields. - Symptom: An error indicating that only stdio processes are supported is displayed when calling an MCP encapsulated with the SSE protocol. Cause: The installed FastGPT version is lower than v4.14.0. Cross-process call adaptation for the SSE protocol is not fully implemented, so the encapsulated MCP cannot be recognized.
How to confirm the configuration is correct
- Manually upload a single special equipment use registration certificate, confirm that the associated inspection report and maintenance record of the device can be recalled during retrieval.
- View the knowledge base synchronization logs, confirm that monthly data synchronization tasks are executed as scheduled with no failed records.
- Adjust the
similaritythreshold, verify that the relevance of recall results meets the matching requirements for supervision documents, with no excessive irrelevant non-supervision documents included. - Check the dataset configuration, confirm that
collectionIdsonly includes dataset IDs related to special equipment supervision, with no irrelevant datasets mixed in.
The values provided here 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 2026-09-28.