What the data for this category looks like
Reference only; verify against the target FastGPT release and deployment documentation.
What constraints these characteristics impose on the citation source and traceability link
The specialized fields and hierarchical structure of rail transit equipment policy data require precise matching of equipment application scope and effective date during citation traceability, to avoid mixing policies across periods and vehicle types. The presence of specialized parameter units requires that traceability results fully retain original markings, without unauthorized modification or omission. Irregular update cycles mean the knowledge base must regularly sync the latest policy documents. Otherwise, traceability results may reference revoked old regulations. The multi-vehicle adaptation document structure requires combining the specific equipment type of the query to filter recalled content during retrieval, preventing irrelevant policies from being included in the citation list.
How to set the configuration
| Configuration Item | Recommended Approach | Basis for This Approach |
|---|---|---|
| Maximum number of citations per request | Top 6 entries | Single rail transit policy document contains lengthy content; excessive citations will exceed the context window, leading to truncated model output |
quoteMaxToken | 8000 token | Matches the average content length of single policy documents, ensuring complete citation of core compliance terms |
responseData | Enabled | Used to return original metadata such as policy document number and issuing authority for citations, supporting traceability verification |
detail | Enabled | Displays specific paragraphs of cited content, facilitating verification of the match between policy clauses and queries |
| Application scope filtering rules | Bound by equipment type | Only recall policy documents matching the rail transit equipment type of the current query |
Three common errors
- Phenomenon: During mixed retrieval, recalled policy documents do not match the current query's rail transit equipment type, and irrelevant content is output. Cause: No citation source filtering rule configured based on equipment application scope, resulting in simultaneous recall of cross-vehicle policies.
- Phenomenon: After configuring citation variables or knowledge base IDs, the output result remains unchanged and an error occurs during application. Cause: Failed to correctly bind access permissions to the rail transit equipment-specific policy knowledge base, or the variable format did not match system requirements.
- Phenomenon: After adjusting the similarity threshold and citation upper limit parameters, the number of recalled citations remains fixed and cannot increase. Cause: Knowledge base chunks are not divided separately by policy document number or effective date, leading to repeated recalls or inability to accurately match query conditions.
How to confirm the configuration is complete
- Initiate a policy query targeting a specific rail transit equipment type, check the citation list in the output result, confirm that each citation includes metadata such as policy document number and issuing authority.
- Switch queries for different types of rail transit equipment, verify that recalled policy documents only contain content within the corresponding application scope.
- Adjust the value of
quoteMaxToken, check whether the total length of cited content changes with the parameter. - Test historical conversation scenarios, confirm that previously generated policy citation records can be correctly retrieved and traced.
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 2026-09-28.