What This Category of Data Looks Like
Data for coal transportation and sales ledgers and reports primarily comes from weighbridge measurement systems, purchase and sales contract management modules, coal quality test reports, and railway/road transport documents. Data is updated per transaction batch or daily. A single ledger document typically includes fields such as coal type identifier, shipment volume, net calorific value of as-received basis, total moisture, buyer and seller entity information, settlement unit price and total amount, and transportation start and end mileage. Units include tons, kcal/kg, yuan, kilometers, and others. Summary report documents integrate multiple transaction records, and include cycle statistics and trend analysis content.
What Constraints Do These Characteristics Impose on Knowledge Base Retrieval and Recall
Coal transportation and sales data is updated at high frequency per transaction batch. This requires knowledge base retrieval to support incremental synchronization to ensure data timeliness. The multi-field document structure with specialized units requires retrieval to precisely match field semantics and unit identifiers, to avoid confusion across coal types or settlement cycles. Single ledger documents have short length, while summary report documents have longer length. Chunking and recall processes must retain the integrity of field associations, to prevent loss of specialized information. Additionally, content related to purchase and sale entities and settlement data must have strict retrieval scope limits, to avoid mixing in irrelevant data.
Configuration Settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Coal transportation and sales ledger fields are specialized and require high precision, to avoid low-match irrelevant data being included in retrieval results |
limit | Top 10 entries | Single ledger data volume is small; excessive recall increases context redundancy and reduces report generation accuracy |
searchMode | Professional semantic matching | Adapts to specialized terms such as coal calorific value and shipment volume, improving semantic matching precision |
usingReRank | Enabled | Re-ranks initially recalled results by relevance to ensure high-match ledger data is prioritized |
datasetSearchUsingExtensionQuery | Enabled | Translates user queries for ledger consolidation into retrieval terms more aligned with coal transportation and sales professional scenarios, covering more relevant documents |
collectionIds | Bind collection IDs for coal transportation and sales ledgers and reports | Restrict retrieval scope to avoid mixing in knowledge base data from other categories |
Three Common Misconfigurations
- The knowledge base management page displays all uploaded documents, but the model returns "No relevant knowledge base content found" during conversation. The cause is that the
collectionIdsbound to the current conversation are not specified as the collection IDs for the coal transportation and sales ledger knowledge base, so the retrieval scope is not limited to the target knowledge base. - The number of chunks after splitting an uploaded summary report document exceeds 3000, and the number of retrieval results during conversation is far lower than expected. The cause is that the
limitparameter was not adjusted, or the ultra-long document was not properly split, resulting in exceeding the system's default chunk limit during retrieval and failing to recall complete data. - Retrieval results include documents not from the coal transportation and sales category, leading to irrelevant content in report generation. The cause is that the
collectionIdswere not configured to restrict the retrieval scope, or the wrong knowledge base collection was bound.
How to Confirm Configuration is Correct
- Enter the knowledge base management page, confirm that the bound
collectionIdspoint to the exclusive collection for coal transportation and sales ledgers and reports. - Initiate a test query, input a scenario-aligned business question such as "2024 Q3 coal shipment volume summary", check if all returned retrieval documents fall under the coal transportation and sales ledger or report category.
- Adjust the value of the
similarityparameter, observe changes in retrieval result matching accuracy to confirm the parameter takes effect. - After enabling
usingReRank, compare retrieval result rankings before and after enabling to confirm the re-ranking function operates normally.
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 2026-09-28.