What this category of data looks like
This category of data comes from state-owned enterprise procurement platforms, public resource trading platforms, supplier databases, contract management systems, and audit issue databases. The update frequency is daily. Documents processed daily include procurement plans, requirement specifications, bidding documents, bid evaluation records, contracts, acceptance documents, and supplier complaint materials. Data fields cover procurement subjects, budget basis, procurement methods, scoring rules, supplier qualifications, performance acceptance information, and other relevant items. The length of individual documents varies widely, ranging from hundreds of characters to tens of thousands of characters.
What constraints do these characteristics impose on the knowledge base retrieval and recall link
Data sources for this category are scattered and updated daily. This requires the retrieval link to support incremental synchronization and associated retrieval of cross-source data. Document types cover multiple categories such as procurement plans and bid evaluation records, with significant structural differences. This necessitates adaptation to text parsing and field matching for different formats. Data fields include core compliance-related information such as procurement methods, budget basis, and scoring rules. Retrieval must accurately locate specific field content to support compliance verification. At the same time, balance must be maintained between internal enterprise procurement rules and state-owned asset supervision requirements. Recall results must cover document entries corresponding to both types of rules, to avoid missing compliance verification basis.
How to configure settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Meets the accuracy requirement for matching compliance-related content in this category, and avoids recalling irrelevant non-compliance documents |
limit | Top 10 entries | Covers the number of relevant documents required for compliance verification across the entire bidding and procurement workflow, while controlling context length |
searchMode | vector | Adapts to the semantic matching needs of multiple types of compliance documents in this category, and follows the semantic retrieval configuration specifications of FastGPT v4.17.0 |
usingReRank | Enabled | Performs secondary ranking on initial recall results, to improve the ranking priority of documents related to high compliance risks |
datasetSearchUsingExtensionQuery | Enabled | Expands the retrieval scope based on compliance keywords from user inquiries, to cover relevant documents such as procurement methods and qualification requirements |
collectionIds | Specified according to the exclusive directory for bidding and procurement compliance | Limits the retrieval scope to only documents exclusive to this category, and avoids mixing irrelevant data from other business modules |
Three common configuration errors
- An
invalid configuration parameter name "hnsw.max_scan_tuples"error occurs during retrieval. This happens when a parameter name not included in the specified configuration list is mistakenly used, and official FastGPT-supported retrieval configuration items are not utilized. - All documents are visible in the knowledge base page, but the model prompts that the knowledge base is empty during conversation. This occurs because the
collectionIdsconfiguration item is not correctly specified, the retrieval scope is not associated with the directory that actually stores documents, or the directory association configuration was not updated synchronously after database migration. - Knowledge base content is not displayed synchronously after upgrading the platform version. This is because the database migration step was not completed, or metadata and vector data of compliance documents were not fully synchronized during the migration process.
How to confirm the configuration is complete
- View the retrieval configuration panel, confirm all configuration items are within the official FastGPT-specified list, with no extra non-list parameters included.
- Initiate a test inquiry for procurement method verification, and check whether the recall results include compliance documents such as corresponding procurement plans and bidding documents.
- Perform an incremental synchronization operation, confirm that daily updated compliance documents can be retrieved and recalled without omission.
- Check the
collectionIdsconfiguration, confirm that only the IDs of the exclusive bidding and procurement compliance directory are included, and no documents from other business directories are mixed in.
The values provided here are common starting points, and should be evaluated 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.