What the data for this category looks like
The data for this category comes from the National Press and Publication Administration’s official service platform, copyright registration system, game package files, and operator materials. Updates are issued irregularly during the declaration cycle for each game. Daily documents include declaration request files, game application forms, copyright certificates, and review reports. Each document typically contains fixed fields such as declaration number, copyright registration number, and review conclusion. Some documents have multiple version iterations, and document formats vary significantly across different sources.
What constraints do these characteristics impose on knowledge base retrieval and recall
Decentralized data sources create large format differences across documents. During retrieval, teams must adapt to the field characteristics of multi-source documents to avoid retrieval failure caused by format incompatibility. Irregular updates mean the knowledge base must support incremental updates to prevent expired approval materials from being mistakenly retrieved. For documents with multiple versions, precise matching of the latest version of declaration materials is required to avoid returning outdated content. Internal tickets often involve professional compliance issues. Retrieval results must be strictly linked to official source data to ensure content compliance and credibility.
How to set configurations
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Game license declaration materials contain a large number of technical terms. This range filters low-relevance results while retaining weakly relevant content related to compliance |
limit | Top 6 results | Internal ticket replies balance comprehensiveness and readability. Too many results exceed the context window limit |
searchMode | Semantic Retrieval | Declaration materials include structured fields and professional expressions. Semantic retrieval better matches vague questions in tickets |
usingReRank | Enabled | Relevance ranking of multi-source documents is easily disrupted by format differences. Re-ranking optimizes the relevance priority of results |
datasetSearchUsingExtensionQuery | Enabled | Internal tickets often contain omitted context. Extended queries supplement implicit fields such as declaration numbers |
collectionIds | Bind exclusive game license declaration knowledge base collection | Isolate documents from other categories to avoid retrieving irrelevant business materials |
Three common configuration errors
- Issue: AI replies do not display knowledge base citation markers, and retrieval results have no associated document sources. Cause: The
collectionIdsparameter is not configured correctly, causing the retrieval scope to cover non-target knowledge base collections. - Issue: The text understanding model list is empty when creating a knowledge base, and the corresponding model cannot be selected. Cause: The API key for the corresponding model is not configured, or the FastGPT version does not support the current model version.
- Issue: WeChat Work conversations cannot trigger knowledge base retrieval, and a connection failure prompt is displayed. Cause: The WeChat Work integration configuration has not been updated, or the corresponding permission configuration has not been completed in FastGPT v4.17.0.
How to confirm the configuration is complete
- Submit a test ticket containing the target declaration number, and check whether the expected knowledge base citation markers appear in the reply.
- View the knowledge base configuration page, and confirm that the
collectionIdsparameter is bound to the exclusive game license declaration knowledge base collection. - Test the sorting of retrieval results after enabling
usingReRank, and verify whether the relevance of the re-ranked results meets business requirements. - Check the model configuration page, confirm that the API key for the text understanding model has been correctly configured, and the model list loads normally.
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 2026-10-19.