What the Data for This Category Looks Like
Petroleum refining business data sources include DCS/APC systems, LIMS laboratory management systems, tank farm metering systems, crude oil evaluation databases, equipment management systems, and energy management systems. Data update frequencies cover three levels: real-time, shift-based, and daily. Daily documents include operating procedures, production plans, material balance sheets, distillate export quality reports, interlock logs, shift handover records, and change management documents. Data fields include crude oil property parameters, unit operation parameters, product quality indicators, and equipment status parameters, with corresponding units such as degrees Celsius, milligrams per kilogram, cubic meters, and others.
Constraints Imposed by These Characteristics on the "Citation Sourcing and Traceability" Link
Data from multiple sources with significantly different update frequencies requires that traceability accurately match data generation times and corresponding source systems to avoid referencing expired or unupdated information. The business logic of multiple continuous units coupled through material and energy networks requires that traceability can associate the complete link of upstream and downstream material flow. Diverse document types and specialized fields require that traceability can distinguish the citation weight of different documents and prioritize core business documents. Highly specialized data fields require that traceability results fully retain professional parameters and avoid truncating critical information.
How to Configure the Settings
This configuration is adapted for FastGPT v4.17.0:
| Configuration Item | Recommended Approach | Rationale |
|---|---|---|
quoteMaxToken | 1200–1500 token | Matches the single-entry content length of refining-specific documents such as distillate export quality reports, to avoid truncating critical parameters |
responseData | Include source system, update time, document type | Meets the multi-dimensional information verification needs during traceability |
detail | Enabled | Supports returning complete cited document fragments and associated metadata |
| Maximum context | 8000–10000 characters | Accommodates business associated data from coupled multiple units |
| Single search limit | Top 6 entries | Covers major material balance and quality report data sources |
| Reranking | Sort by source system priority | Prioritizes core business data from DCS and LIMS |
Three Common Misconfigurations
- Symptom: Citation file lists are automatically appended to the end of conversation outputs and cannot be hidden. Cause: The output configuration for
responseDatawas not adjusted, retaining the default citation display logic. - Symptom: Citation fragments are truncated, and critical quality parameters such as distillation range and sulfur content are lost. Cause: The value of
quoteMaxTokenis smaller than the single-entry content length of professional documents, resulting in forced content truncation. - Symptom: Traceability results fail to match evaluation data for the corresponding crude oil batch. Cause: Search results were not sorted by data source priority, and non-core document sources were prioritized.
How to Verify Proper Configuration
- Run a query covering crude oil properties and unit load, and confirm returned results include source system, update time, and document type information.
- Test a long-text query, and confirm citation fragments are not truncated and fully contain professional parameter content.
- View the configured value of
quoteMaxToken, and verify it matches the average length of business documents. - Run a multi-unit associated query, and confirm traceability results associate upstream and downstream material and energy flow data.
The values provided here are common starting points and should be measured against the reader's own samples.