What this category’s data looks like
The data for this category mainly comes from state-owned enterprise procurement centers, tendering offices, business demand departments, audit and discipline inspection departments, tendering agencies, and other relevant entities. Sources include enterprise procurement platforms, public resource trading platforms, supplier databases, contract management systems, audit problem databases, and more. Daily documents cover procurement plans, demand specifications, tendering documents, bid evaluation records, contracts, acceptance receipts, and supplier complaint materials. Data is updated on a daily basis. Most documents include core compliance fields such as procurement methods, budget basis, scoring methods, and supplier qualification requirements. Formats combine structured tables and unstructured text.
What constraints these characteristics impose on the knowledge base retrieval and recall link
The daily data update feature requires configuring incremental synchronization rules to prevent full refreshes from consuming computing resources. Mixed structured and unstructured document formats require adapting chunking logic to preserve the integrity of core compliance fields. For retrieval requests involving sensitive compliance content such as procurement methods and budget basis, high-match compliance basis documents should be prioritized for recall. Scenarios with multi-department collaborative use require the knowledge base to support collaborative sharing modes. The recall upper limit must also be adjusted based on data fluctuations during the procurement cycle to adapt to request pressure across different stages. Additionally, compliance-focused retrieval must avoid returning irrelevant content, so a reasonable matching threshold must be set.
How to configure
Reference only; verify against the target FastGPT release and deployment documentation.
| Configuration Item | Recommended Value Range | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches core fields related to tendering and procurement compliance, filters low-relevance non-compliance documents |
limit | Top 8–12 entries | Covers multiple types of compliance references required for single draft document creation, avoids context overload |
searchMode | hybrid | Adapts to mixed structured and unstructured document formats, balances keyword and semantic matching |
usingReRank | Enabled | Re-ranks recall results to prioritize matches for core compliance items such as procurement methods and scoring methods |
datasetSearchUsingExtensionQuery | Enabled | Automatically expands query terms, supplements compliance-related synonyms or related expressions |
collectionIds | Bind specified collections by business department | Isolates compliance data for different procurement scenarios, avoids cross-department data interference |
Three common errors
- Phenomenon: A large number of general documents unrelated to tendering and procurement compliance appear in retrieval results. Cause: The
similaritythreshold is not set, or the threshold is set too low, failing to filter low-match content. - Phenomenon: The knowledge base cannot split documents containing "\n" strings according to actual line breaks. Cause: Correct chunking rules are not configured, and actual line breaks in text are not recognized.
- Phenomenon: The target knowledge base ID cannot be automatically retrieved in the workflow. Cause: The knowledge base ID exposure function is not enabled in system settings, or the correct node configuration parameters are not used.
How to confirm the configuration is complete
- Initiate a query that includes procurement methods and budget basis, and verify that the recall results include corresponding compliance documents.
- View the knowledge base synchronization log to confirm that the daily incremental update task runs normally.
- Invite multiple test accounts to access the same knowledge base, and confirm that content can be retrieved and edited normally.
- Call the knowledge base retrieval node in the workflow, and confirm that the target knowledge base ID can be automatically obtained.
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.