What This Category of Data Entails
Data for this category primarily originates from the low-income population dynamic monitoring platform, resident family economic status verification system, and cross-department shared data from public security, human resources and social security, medical insurance, housing and urban-rural development, and tax authorities. The update frequency is monthly. Daily generated documents include minimum living security application forms, family economic status verification reports, household survey forms, democratic appraisal records, public notification materials, and assistance approval forms. Data fields focus on family economic status, member relationships, and assistance eligibility, with no additional custom statistical values.
Constraints Imposed by These Characteristics on the Citation Sources and Traceability Process
Cross-department shared data sources have format differences. This requires the retrieval link to anchor core fields such as family income and member relationships for precise matching, to avoid citation deviations caused by format incompatibility. The monthly update frequency requires the knowledge base synchronization cycle to align with the data update rhythm, to ensure the timeliness of retrieved content. Document types are fixed and focused on assistance eligibility. The traceability link must clearly mark the specific system and update node of the data source, to ensure cited content is traceable and avoid risks of incorrect or missed assistance approvals.
How to Configure
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting Range | Rationale |
|---|---|---|
| Similarity Threshold | 0.35–0.45 | 0.35–0.45 |
quoteMaxToken | 800–1200 characters | Covers the core content of a single verification report or application form, and avoids truncation of key supporting information |
| Single Search Limit | Top 6 results | Controls the number of retrieval results to avoid redundant information interference, and adapts to the limited volume of cross-department data |
responseData | Enable source field labeling | Clearly displays the system and update time from which the data originates, to meet traceability requirements |
detail | Enable full citation | Retains the original fields of the cited content to avoid information loss |
| Maximum Context | Calibrated based on actual testing | Adapts to the total knowledge base capacity and avoids exceeding the model's context limit |
Three Common Misconfigurations
- Setting the similarity threshold to 0.4 still fails to match core data in family economic status verification reports. This occurs because the threshold is not adjusted for structured fields, and a general threshold cannot meet the precise matching needs of civil affairs documents.
- After uploading 100 5kb files to the knowledge base, half of the files show training exceptions. This occurs because format differences in cross-department shared data cause chunk parsing failures, and the structured document parsing mode is not enabled.
- Unable to set the temperature parameter when selecting a variable reference large model in the AI conversation component. This occurs because the large model exclusive configuration page is not accessed, and configuration access is missing when calling directly in the conversation component.
How to Verify Successful Configuration
- Generate a draft request for a minimum living security application form, and check whether the cited content includes core fields such as family income and member relationships.
- View the knowledge base synchronization log to confirm that the monthly update task has been completed as scheduled.
- Open the citation traceability panel to check whether each cited content is labeled with the source system and update time.
- Adjust the similarity threshold and verify whether the matching accuracy of retrieval results meets business requirements.
The values provided are common starting points and should be measured against the user'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.