What this category of data looks like
Daily documents for this category include batching lists, smelting records, electrolytic cell condition records, laboratory reports, anode and cathode quality records, and environmental protection ledgers. Data sources cover concentrate and auxiliary material analysis, DCS systems, cell voltage and current monitoring, laboratories, and energy and environmental monitoring equipment. Updates are triggered in real time for each batch of raw materials, each furnace, or each cell. Individual documents include fields such as cell number, current and voltage values, impurity content, metal recovery rate, byproduct disposal, and more. Units include amperes, volts, grams per ton, cubic meters, and others.
What constraints do these characteristics impose on the knowledge base retrieval and recall link
Real-time updated multi-source data requires retrieval to quickly match the latest production records, avoiding the use of outdated data for ledger summary. The mixed scenario of multiple types of documents requires retrieval to support both structured field matching and unstructured content recall, adapting to the query needs of different ledgers. Complex component balance and furnace condition control requirements require that recalled results accurately associate production data for specific cell numbers and batches, avoiding irrelevant information from interfering with accounting. Strict consequences of errors require retrieval results to have a relatively high relevance threshold, preventing low-quality content from affecting the accuracy of final reports.
How to set the configurations
| Configuration Item | Suggested Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Filter low-correlation historical records, avoid irrelevant cell condition data interfering with metal balance accounting |
limit | Top 10–15 entries | The volume of smelting ledger data per batch is limited; excessive recall will exceed the context window and affect report generation efficiency |
searchMode | mixedRecall | Balance the matching needs of structured fields (such as cell number, batch number) and unstructured furnace condition descriptions |
usingReRank | Enabled | Prioritize recalling the latest records related to the current production cell number, improve the ranking weight of key information |
datasetSearchUsingExtensionQuery | Enabled | Expand the user's ledger summary request into a related query matching batching lists and laboratory reports, covering more comprehensive production data |
collectionIds | Bind the knowledge base collections corresponding to the smelting workshop | Isolate ledger data from different production lines, avoid cross-shop records interfering with the summary work of the current batch |
Three common misconfigurations
- Phenomenon: Retrieval results include electrolytic cell records from unrelated workshops, and some fields are empty. Cause:
collectionIdsis not configured correctly, and the exclusive knowledge base collection of the corresponding production line is not bound. - Phenomenon: Retrieval request times out, returns a service timeout status code. Cause: The
limitparameter is not adjusted, and excessive recall of historical ledger records causes parsing and context splicing timeouts. - Phenomenon: Unable to filter recalled results by cell number or batch, and the tag filtering function is unresponsive. Cause: The tag entry function of the knowledge base is not enabled, or the corresponding filtering rules are not configured in the retrieval module.
How to confirm the configuration is correct
- Upload one recent electrolytic cell condition record, initiate a test request containing "view yesterday's data for cell #5", check whether the recalled results only include production records for cell #5.
- View the retrieval logs of FastGPT v4.17.0, confirm that the switch status and configuration of
usingReRankanddatasetSearchUsingExtensionQuerymatch the set values. - Test adjusting the
similaritythreshold, observe changes in the relevance of recalled results, confirm that the value meets the filtering needs of the current business. - Check the binding list of knowledge base collections, confirm that only ledger documents such as batching lists and smelting records for the current smelting workshop are included.
The values provided here are common starting points and should be evaluated against operational samples from the target scenario.