What This Category of Data Looks Like
Data for this category comes from enterprise internal control systems, ERP systems, financial shared services systems, audit management systems, and procurement contract project systems. Update frequency is annual. Daily documents include internal control self-assessment reports, audit notices, audit working papers, issue lists, rectification ledgers, and rectification closure certificates. Most documents contain fields such as business sample numbers, audit issue descriptions, rectification deadlines, and closure statuses. There is no unified fixed unit. Content centers on business records related to post-event inspection and rectification closure. This category focuses on issue discovery and rectification closure as core goals. It differs from legal compliance scenarios, with greater emphasis on business records for post-event inspection and evidence sampling.
What Constraints Do These Characteristics Impose on the Knowledge Base Retrieval and Recall Link
The annual update frequency of the data means knowledge base synchronization does not require high-frequency triggers. It can be configured as annual full incremental updates combined with small-scale daily synchronization. Document types include structured ledgers and unstructured working papers. Differences in fields across documents require retrieval to balance structured matching and semantic recall. Data from multiple systems is stored dispersedly. The associated dataset scope must be clearly specified to prevent irrelevant cross-category data from being included. Content focuses on audit issues and rectification closure. Retrieval must accurately match policy basis and business scenarios to avoid recall results deviating from audit compliance requirements. Long documents account for a large share. Single-block recall length must be limited to avoid context overflow affecting model understanding.
Configuration Settings
The following configurations are implemented using the knowledge base retrieval feature of FastGPT v4.17.0.
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
similarity | 0.75–0.85 | Meets the rigor requirements of audit documents, filters low-match irrelevant policy content |
limit | Top 8–12 results | Covers key information across multiple document types such as rectification ledgers and audit working papers, avoids retrieving excessive redundant content |
searchMode | Semantic + keyword hybrid retrieval | Balances keyword matching for policy clauses and semantic association for business scenarios, adapts to the precise retrieval needs of audit scenarios |
usingReRank | Enabled | Performs secondary ranking on initial recall results, improves the display priority of high-match policy basis |
datasetSearchUsingExtensionQuery | Enabled | Generates extended query terms based on user questions, covers synonyms and near-synonyms of policy basis, adapts to professional terminology in policy texts |
collectionIds | Configured for audit and rectification exclusive datasets | Limits retrieval scope to exclusive datasets related to internal control and audit, prevents mixing of business data from other categories |
Three Common Misconfigurations
- Retrieval results return a large number of non-audit related business documents. This occurs when
collectionIdsis not configured to limit the retrieval dataset, allowing redundant data from other business systems to be included. - Retrieval times out or returns empty results. This happens when
limitis not set appropriately, with too many long documents retrieved causing context overflow, or when thesimilaritythreshold is set too high, resulting in no matching content. - Recalled policy basis does not match the current audit issue. This is caused by not enabling
usingReRankordatasetSearchUsingExtensionQuery, which prevents accurate matching of professional terminology and scenario association.
How to Confirm Proper Configuration
- Manually enter a typical audit question, check that retrieval results only include internal control and audit related policies and documents.
- View the knowledge base synchronization logs to confirm the update frequency matches the annual synchronization configuration.
- Test retrieval across different types of audit documents to verify the recall effect of semantic and keyword hybrid retrieval.
- Check the dataset association configuration to confirm only exclusive data source collections for this category are included.
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 2026-09-28.