What this category of data looks like
This category of data primarily originates from daily business documents of central state-owned enterprise procurement centers, tendering offices, business requirement departments, audit and discipline inspection departments, and tendering agencies. Documents include procurement plans, requirement specifications, tender documents, bid evaluation records, contracts, acceptance receipts, supplier complaint materials, and more. Data sources cover enterprise procurement platforms, public resource trading platforms, supplier databases, contract management systems, and audit issue databases. The update frequency is daily. Data fields include procurement methods, budget basis, scoring methods, supplier qualification requirements, performance acceptance standards, and others. No unified fixed units exist; values shift based on specific business scenarios.
What constraints do these characteristics impose on the knowledge base retrieval and recall link
Data sources are scattered across multiple business systems. Unified cross-source data recall and permission verification must be supported. The daily update requirement demands that the retrieval pipeline supports incremental synchronization and real-time verification. This ensures the latest policies and procurement documents can be retrieved quickly. Daily documents use diverse formats, including structured tables, PDFs, Word files, and others. Multi-format file chunking and precise recall must be supported. Business scenarios focus on compliance verification. Retrieval results must accurately match policy and regulatory clauses. Irrelevant content must not disrupt query results. The system must also support traceability requests from audit and discipline inspection departments. Retrieval results must link back to original files. This enables retrospective verification for compliance reviews.
Configuration Settings
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the precision requirement for tendering and procurement policy clauses, avoids low-relevance results being included or core clauses being missed |
limit | Top 10 entries | Covers the core clause count for common compliance checks, balances retrieval efficiency and result completeness |
searchMode | keyword+vector | Combines keyword matching for explicit policy clause phrasing and vector recall for semantically related content, fits the diverse needs of compliance queries |
usingReRank | Enabled | Re-ranks initial recall results, raises the display priority of highly compliance-relevant content |
datasetSearchUsingExtensionQuery | Enabled | Automatically expands query terms to cover synonymous and related phrasing of policy clauses, improves recall coverage |
collectionIds | Bound to tendering and procurement compliance business modules | Only retrieves corresponding knowledge base collections, eliminates interference from irrelevant business data |
Three Common Mistakes
- Symptom: After uploading a PDF file, no valid chunks are generated, and no corresponding content appears during retrieval. Cause: The call path for pdf-marker is not configured correctly, or there is a network connectivity issue between FastGPT and pdf-marker.
- Symptom: An access denied error is returned during retrieval. Cause: The current query role is not configured with access permissions for the corresponding knowledge base collection, or the
collectionIdsparameter binds collections outside the permitted scope. - Symptom: Clicking a retrieval result fails to open the online preview details of the original file. Cause: The knowledge base's original file traceability configuration is not enabled, or the storage path metadata of the original file was not retained during chunking.
How to Confirm Configuration is Complete
- Run a single compliance query, such as "Applicable conditions for open tendering". Check if the number and matching degree of returned results meet expectations.
- Click the associated link of any retrieval result, verify that the online preview page of the original file can be opened.
- View the knowledge base's synchronization logs, confirm that daily updated procurement documents have been successfully included in the retrieval scope.
- Switch between different business roles, verify that the permission configuration takes effect. Roles without permissions cannot retrieve content from the corresponding collections.
The values given 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.