What Data for This Category Looks Like
Daily documents for this category include growth experiment plans, user segmentation tables, outreach copy, and retention analysis reports. Data sources include user behavior logs, push systems, experiment platforms, and CRM systems. Update frequency is mostly daily or weekly. Common data fields include user segmentation tags, outreach channels, experiment variables, and retention metrics. There are no uniform fixed units, and adjustments are needed based on specific document scenarios.
Constraints Imposed by These Characteristics on Knowledge Base Retrieval and Recall
Documents in this category have diverse formats, including structured tables, semi-structured experiment plans, and plain-text outreach copy. This requires the knowledge base retrieval to adapt to recall logic for multi-format content. Data is updated daily or weekly, so an incremental synchronization mechanism must be supported to ensure recalled content aligns with the latest operational data. Fields from internal system sources are scattered, so filtering retrieval by business collections must be supported to accurately match scenario needs such as user segmentation and outreach rhythm. Content involving user behavior and outreach information requires strict permission control. The relevance of recall results directly affects operational effects, and incorrect content must be avoided to prevent user dissatisfaction or wasted budget.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
collectionIds | Bind the dataset ID corresponding to the current operational scenario, such as user segmentation dataset, outreach copy dataset | Implement filtering retrieval by business collection to accurately recall exclusive documents required for the category |
similarity | 0.72–0.88 | Balance the relevance and coverage of recalled content, adapting to retrieval matching for multi-format documents |
limit | Top 3–6 entries | Control the number of single recall results, adapting to context window limits in operational scenarios |
usingReRank | Enabled | Optimize the sorting logic of recall results, improving matching accuracy for structured tables and experiment report content |
datasetSearchUsingExtensionQuery | Enabled | Expand relevant retrieval terms based on the original question, adapting to fuzzy matching needs in growth operation scenarios |
searchMode | Semantic retrieval | Adapt to scenarios where most category documents are business logic descriptions, improving retrieval accuracy |
Three Common Configuration Mistakes
- Phenomenon: Retrieval results include documents unrelated to the current operational scenario, and collection filtering is not implemented. Cause: The
collectionIdsparameter is not configured correctly, or the bound dataset ID does not match the target business collection. - Phenomenon: Retrieval details only display a single original question, and expanded retrieval terms are not shown. Cause: The
datasetSearchUsingExtensionQueryconfiguration is not enabled, or the FastGPT version is v4.8.10 or earlier. The display logic has been optimized in versions v4.17.0 and above. - Phenomenon: Retrieved content cannot be directly used for user segmentation or experiment attribution analysis. Cause: The
usingReRankconfiguration is not enabled, and the recall result sorting logic does not adapt to field matching requirements for structured documents.
How to Verify Successful Configuration
- View the retrieval result list, confirm only documents from the datasets corresponding to the bound
collectionIdsare included, to verify the collection filtering configuration is active. - Check the retrieval details page, confirm whether expanded retrieval terms are displayed, to verify the
datasetSearchUsingExtensionQueryconfiguration is enabled. - Review the sorting of recall results, confirm content with higher relevance is displayed first, to verify the
usingReRankconfiguration is active. - Input structured user segmentation-related questions, confirm returned results include field information from user segmentation tables, to verify the multi-format document retrieval adaptation configuration is active.
The values provided are common starting points and should be measured against samples from the actual deployment.
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.