What the data for this category looks like
The core data for this category comes from employee directories, email gateways, internal training platforms, and phishing drill click statistics. Daily produced documents include training courseware, drill plans, phishing email templates, and drill result reports. Fields cover drill audience scope, email sending records, employee click feedback, and rectification records. Data updates occur quarterly or annually. Individual document sizes range from dozens to hundreds of pages. The core associated fields are employee ID and drill batch identifier.
What constraints these characteristics impose on citation sources and traceability
Dispersed data sources and reliance on employee ID and drill batch for associated fields mean multiple system identity identifiers must be matched during traceability. Unified metadata from general knowledge bases cannot be directly reused. Low update frequencies require cited document versions to be strictly bound to drill batches, to avoid cross-batch confusion. Diverse document types need dedicated extraction rules for formats such as training courseware and email templates. Without these rules, field missing will occur. Traceability for drill result reports requires association with click statistics data. Without cross-source association logic configured, actual employee participation behavior cannot be verified during post-drill reviews.
How to set the configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 token | Documents in this category are mostly training courseware and drill reports, with long single citation contents, requiring coverage of complete drill plan or result fragments |
responseData | Enable document title, drill batch, and employee ID fields | Binding drill batches and identity identifiers ensures that original data for the corresponding drill can be quickly located during traceability |
detail | Enable full traceability mode | Traceability for phishing drills requires original click statistics records; summaries alone cannot meet audit requirements |
| Cross-source association matching rules | Match using both employee ID and drill batch fields | Data sources are dispersed, so both identity and batch must be matched to accurately associate data from email gateways and training platforms |
| Document format adaptation | Configure extraction rules separately for email templates and training courseware | Large format differences exist across documents in this category; generic extraction will lose critical metadata |
| Citation display toggle | Enable citation display on the official chat page | Debug pages can display citations but official pages may not; this configuration must be enabled to ensure traceability information is visible |
Three common configuration errors
- Issue: Citation sources are not displayed on the official chat page, but show correctly on the debug page. Cause: The official page citation display configuration is not enabled, and only the debug mode toggle is retained.
- Issue: Extracted traceability content only includes text fragments, and does not include metadata such as drill batch and employee ID. Cause: The
responseDatametadata extraction parameters are not configured. Only core text content is extracted, with no associated fields bound. - Issue: Cross-source association traceability results are empty, and email gateway click records cannot be matched. Cause: Only a single employee ID is used as the matching field, and drill batch is not bound at the same time. This leads to cross-batch data confusion and inaccurate matching.
How to confirm the configuration is correct
- Enter the chat interface released in FastGPT v4.17.0, initiate a query related to phishing drills, and check if cited document titles and batch information are displayed below the reply.
- Export the traceability log, and verify that extracted metadata includes employee ID, drill batch, and click statistics associated fields.
- Initiate queries for different document types such as training courseware and email templates, and confirm that extracted content formats meet expectations with no critical fields missing.
- Adjust the
quoteMaxTokenvalue, and verify that cited content length conforms to configuration requirements, with no forced truncation or content overflow.
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 2026-10-19.