Applicable to FastGPT v4.17.0
What this category's data looks like
The typical entities in this category are slaughterhouses, cooked meat and frozen prepared food manufacturers. Daily generated documents include animal quarantine certificates, slaughter and cutting records, ingredient marinating records, thermal processing records, cold chain temperature records, microbial test reports, and other related documents. Data sources cover animal quarantine information, supplier batch ledgers, production traceability systems, temperature collection equipment, metal detection equipment, and laboratory test results. Data update frequency follows production batches and shifts; continuous temperature control data is updated in real time. Fields include unique identifiers such as batch number, temperature value, test time, and production workshop.
What constraints do these characteristics impose on the "citation source and traceability" link?
Batch-updated production data requires traceability to precisely match batch numbers, to avoid traceability errors caused by cross-batch citations. The timeliness of real-time temperature control data requires retrieval to be associated with time intervals, to ensure that referenced temperature records correspond to production periods. Document classification based on raw and cooked product zones requires traceability to distinguish production scenarios, to avoid misquoting documents related to cross-contamination. The need to integrate multi-source data requires traceability to associate common identifiers of different document types, such as batch numbers and quarantine numbers. The severity of potential errors requires traceability information to be complete and accurate, and key test or quarantine records must not be omitted.
How to configure the settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
quoteMaxToken | 800–1200 token | Matches the character length of single category-specific documents (such as quarantine certificates, cold chain temperature control records) to avoid truncating key traceability fields |
responseData | Include complete citation identifiers | Ensures returned cited content includes unique information such as batch number, test time, and production workshop, to support precise traceability |
detail | Enable detailed traceability mode | Displays specific source nodes of cited documents, which meets production traceability requirements for meat product processing |
| Single recall limit | Top 8 retrieval results | Covers multiple types of documents for a single production batch (slaughter records, thermal processing records, microbial test reports) to ensure comprehensive traceability information |
| Retrieval timeout | 600 seconds | Adapts to documents that may be generated with delay such as laboratory test reports, to avoid interruptions during retrieval |
Three common configuration errors
- Garbled characters appear in citation symbols during conversation output, which are finally displayed as standard quotation marks. The cause is that the escape rules for citation identifiers are not correctly configured, causing special characters to be abnormally processed during output.
- When there is no matching content in the knowledge base, irrelevant document citations are still returned. The cause is that the auto-completion logic for no matching results is not disabled, or the recall threshold is set too low, leading to accidental recall of documents from unrelated categories.
- The conversation interface does not return cite citation IDs. The cause is that the complete citation return switch for the
responseDataconfiguration item is not enabled, causing the interface to only return response content without carrying traceability identifiers.
How to confirm the configuration is correct
- Initiate a process query that includes a specific production batch number, check whether the citation at the end of the response includes document information corresponding to that batch number.
- Trigger a query scenario where there is no matching content in the knowledge base, confirm that the system does not return citations of irrelevant documents.
- Call the conversation interface, check whether the returned data includes complete citation traceability fields.
- View the output format of citation symbols, confirm that no garbled characters or abnormal conversion occur.
The values provided are common starting points and should be measured against the reader's own samples.