What Data Looks Like for This Category
Core data for precast concrete components derives from BIM detailed design, rebar and concrete batch ledgers, mold table and curing kiln operation logs, laboratory test reports, and product QR code traceability systems. Daily documents include component detailed drawings, rebar cutting lists, mold inspection forms, pouring and vibration records, curing cycle records, demolding and factory inspection reports. Data updates run per individual component, each production batch, and each kiln run. Fields include component number, rebar diameter, curing temperature, demolding strength, and more. Units include millimeters, degrees Celsius, megapascals, and other common engineering units.
What Constraints Do These Characteristics Impose on Knowledge Base Retrieval and Recall?
Documents for precast concrete components are mostly structured engineering forms and professional drawings, with data linked to specific component numbers and production batches. Retrieval therefore requires precise matching of field information to avoid irrelevant results from broad matching. High-frequency updates at the individual component and batch level require the knowledge base to synchronize the latest inspection and production records regularly, preventing expired data from being used for audits. At the same time, audit results for this category directly affect on-site installation and structural safety. Retrieval scope must be strictly limited to uploaded internal documents, and recalled results must include document section titles to clarify information sources and avoid mixing parameters from different components.
Configuration Settings
The following configurations are based on FastGPT v4.17.0, adapted for the supplier and process quality audit scenario for precast concrete components:
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75-0.85 | Precast concrete components have high parameter precision requirements. This range filters irrelevant documents with insufficient matching scores to avoid incorrect audit results |
limit | Top 8 entries | The number of associated documents for a single component batch is limited. Too many results will exceed the context window and reduce the accuracy of AI responses |
searchMode | mixedRecall | Balances semantic matching for engineering drawings and precise matching for keywords such as component numbers and batch numbers, covering multi-dimensional information required for audits |
usingReRank | Enabled | Re-sorts initial recalled results by relevance to prioritize documents that best match the current audit component |
datasetSearchUsingExtensionQuery | Disabled | Strictly limits AI responses to only use uploaded knowledge base documents, preventing extended queries from introducing external irrelevant content |
collectionIds | Bind to the current plant's precast concrete component document collection | Isolates documents from different projects by production batch, avoids cross-project data confusion, and ensures accurate audit scope |
Three Common Mistakes
- Issue: Retrieval results include general engineering knowledge outside the knowledge base. Cause: The
datasetSearchUsingExtensionQueryparameter is enabled, which automatically expands the query scope and causes the AI to introduce external irrelevant information. - Issue: Retrieval results do not include document section title information. Cause: The section title extraction configuration during document parsing is not enabled, or the association between titles and main text is not retained during chunking, so recalled results cannot clearly identify information sources.
- Issue: Attempting to download an uploaded rebar cutting list document via "View Original Content" returns an inaccessible link. Cause: FastGPT's file hosting domain is not configured correctly, so the generated download link points to an incorrect internal address and cannot access the original document externally.
How to Verify Correct Configuration
- Navigate to the FastGPT knowledge base collection management page and check if the
collectionIdsparameter is bound to the current plant's precast concrete component document collection. - Submit a targeted test query, such as "Demolding strength requirements for C30 precast beams for Project XX", and check if returned results only include uploaded curing records and factory inspection documents, with no external knowledge included.
- Review the context of retrieval results to confirm they include document section title information, such as "3.2 Curing Records", to ensure clear information sources.
- Click "View Original Content" in retrieval results to confirm the download link works properly and points to the uploaded original precast concrete component document.
The values provided here are common starting points and should be measured against the reader's own samples.