What the data for this category looks like
The core data for this category comes from backup systems, database logs, object storage, and disaster recovery platforms. Four document types are regularly generated: backup policies, recovery drill reports, disaster recovery switchover plans, and data verification records. Data update frequencies fall into three categories: daily, monthly, and quarterly. Backup policies and data verification records are updated more frequently. Recovery drill reports and disaster recovery switchover plans are updated according to actual drill cycles. Documents include fields such as backup path, recovery steps, verification results, and drill execution time. Field units match specific business scenarios, with no unified fixed format.
What constraints these characteristics impose on knowledge base retrieval and recall
The data characteristics of this category impose multiple constraints on knowledge base retrieval and recall. Dispersed multi-source data requires retrieval tools to support associated recall across backup systems, database logs, object storage, and disaster recovery platforms, to avoid data silos. Uneven update frequencies require configuring incremental update mechanisms to ensure retrieval results only include the latest backup policies, verification records, and drill documents. Diverse document structures require setting differentiated retrieval weights for different document types, to prioritize results that match core fields. Severe potential consequences require strictly controlling the accuracy of recall results, to avoid invoking incorrect recovery steps or verification logic that disrupts business continuity.
Configuration guidelines
This document is written based on FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Matches the semantics of core documents such as backup policies and recovery drill reports, filters low-relevance redundant content |
limit | Top 8 entries | Covers key information across multiple document types including recovery steps and data verification records, prevents excessive single-recall content from exceeding context limits |
searchMode | semantic + fulltext | Balances semantic matching and keyword retrieval, accurately recalls documents that include backup paths and recovery timestamps |
usingReRank | Enabled | Re-ranks initial recall results, prioritizes content strongly related to disaster recovery drills and recovery operations |
datasetSearchUsingExtensionQuery | Enabled | Expands retrieval terms based on user questions, matches implicit configuration items and verification rules in backup policies |
collectionIds | Specify dedicated dataset ID for disaster recovery drills | Isolates knowledge base content from other categories, prevents recall of irrelevant general operation and maintenance documents |
Three common configuration mistakes
- Phenomenon: "Connection error" is returned during retrieval. Cause: The data source connection for the dedicated disaster recovery dataset was not configured correctly, making it impossible to access raw data from backup systems, database logs, or disaster recovery platforms.
- Phenomenon: The number of single-recall results is insufficient to cover complete recovery steps or verification records. Cause: The
limitvalue was set too low, or thesimilaritythreshold was set too high, filtering out some relevant but slightly lower-matching documents. - Phenomenon: Unable to confirm whether output content originates from backup recovery and disaster recovery drill-related knowledge bases. Cause: Content provenance functionality was not enabled, or a dedicated dataset ID was not specified, leading to recall content mixed with operation and maintenance documents from other categories.
How to verify correct configuration
- Run a test retrieval, input a typical question for this category, and verify whether recall results include dedicated documents such as backup policies and recovery drill reports.
- View retrieval logs, confirm that extended retrieval and re-ranking functions were triggered, and that core fields such as backup paths and recovery timestamps were matched.
- Check dataset configuration, confirm that only the dedicated dataset ID for disaster recovery drills was specified, with no knowledge base content from other categories mixed in.
- Validate exception scenarios, input simulated data source access issues, and confirm that corresponding error prompts are returned pointing to configuration steps.
The values provided 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 2026-10-19.