What this category of data includes
The core data for this category originates from flow, pressure, temperature, and analytical instrument manufacturers, as well as metrology laboratories. Daily documents include assembly and commissioning sheets, calibration procedures, original calibration records, error curves, metrology certificates, and software version logs. Data is sourced from design documents, standard instrument and environmental monitoring data, calibration software, laboratory information systems, and customer range requirements. Updates occur after each calibration for every individual device. Key information in these documents includes traceability chains, error values, metrology validity periods, and software version numbers.
Constraints for Knowledge Base Retrieval and Recall Workflows
This category of data has a high update frequency, with new data generated after each calibration of every device. This requires the knowledge base retrieval and recall workflow to support incremental updates, avoiding the time cost of full reconstruction. Documents contain strongly correlated fields such as traceability chains, error curves, and metrology certificate numbers. Retrieval must accurately match business-related fields, rather than relying solely on general semantic matching. A single calibration record is associated with multiple attached documents. Associated context must be retained during recall. Data errors will directly lead to failed metrology audits, so retrieval result accuracy has stricter requirements.
Configuration Guidelines
This page’s configuration guidelines are based on FastGPT v4.17.0.
| Configuration Item | Recommended Range | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Documents in this category require precise matching of calibration procedures and error data. A threshold that is too low will introduce irrelevant results, while a threshold that is too high will miss valid content. Adjust based on actual testing and calibration. |
limit | Top 8–12 results | The number of calibration-related documents for a single device is concentrated. Too many results will exceed the context window, while too few will fail to cover all associated information. |
searchMode | mixedRecall | Balances semantic matching and precise field recall, adapting to retrieval needs for structured fields such as traceability chains and certificate numbers. |
usingReRank | Enabled | This category of documents contains a large amount of numerical error data. Reranking can filter results that are semantically similar but have excessive numerical deviations. |
datasetSearchUsingExtensionQuery | Enabled | Can automatically supplement associated query terms such as calibration cycles and range requirements based on user inspection questions, improving recall coverage. |
collectionIds | Configured by device model group | This category manages calibration data by device model. Limiting the retrieval scope by group reduces invalid recall. |
Three Common Configuration Errors
- Symptom: Knowledge base search takes longer than 40 seconds, and the interface returns a timeout status code. Cause:
collectionIdsis not configured by device model group, and the retrieval scope covers all calibration documents, resulting in excessive data volume traversed during vector recall. - Symptom: Users can download original calibration records and metrology certificates via the knowledge base export interface. Cause: Document download permissions are not disabled in the dataset configuration, and access restrictions are not set for highly sensitive metrology documents.
- Symptom: CSV-format QA imports fail when the Q field content exceeds 8000 characters. Cause: The knowledge base’s single QA character limit configuration is not adjusted, and the default limit is retained without adapting to the import needs of long-text calibration procedures for this category.
How to Confirm Successful Configuration
- A retrieval for an inspection question related to a single device is performed. Returned results are checked for inclusion of the device’s calibration procedures and error curves, and the
similarityparameter is adjusted to optimize matching accuracy. - Retrieval time logs are reviewed, latency is confirmed to meet business expectations, and the
limitparameter orcollectionIdsgroup configuration is adjusted to optimize retrieval speed. - A single long-text calibration procedure is imported for testing. No character limit errors are confirmed, and the corresponding configuration item is adjusted to adapt to long-text import needs.
- Dataset permission settings are checked. Sensitive documents are confirmed to be unavailable for download or export by regular users, and access restriction configurations are verified to take effect.
- Retrieval of the latest metrology calibration specification content is performed. Online search configurations are confirmed to take effect, and results from public standard data sources are returned.
The values provided here are common starting points and should be measured against local operational samples.