What the data for this category looks like
Data for rural commercial bank marketing content comes from three main sources: internal business systems, offline event materials, and customer tiered operation records. The update schedule for this data adjusts based on business type:
- Product prospectuses are updated weekly
- Credit policy documents are revised quarterly
- Offline event promotional materials are added as needed
Document structures include both structured and unstructured formats. Structured data includes fields such as product code, minimum investment amount, and term, with units of yuan and months. Unstructured data includes event scripts, promotional copy, and poster description text. All data is stored on the bank's internal intranet server. No direct connection to external public data sources is established.
What constraints these characteristics impose on deployment and upgrade
The mixed structure of rural commercial bank marketing content and its intranet storage location create specific constraints for deployment. Structured fields must support custom mapping to ensure accurate matching of product parameters during retrieval. Frequently updated content requires scheduled synchronization tasks to be configured during deployment, to avoid data lag. Intranet storage requires deployment environments to support intranet access permission configuration, to block unauthorized external access.
The upgrade process must be compatible with existing intranet data synchronization links, to prevent data synchronization interruptions from version updates. Backup of in-bank marketing content parsing rules must be completed in advance, to ensure rapid configuration recovery after upgrade.
How to set configurations
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
UPLOAD_FILE_MAX_SIZE | 1000 MB | Rural commercial bank marketing materials often include high-definition posters and complete credit policy compilations, resulting in large individual file sizes |
PARSE_FILE_TIMEOUT_SECONDS | 600 seconds | Parsing long documents such as annual credit policy compilations takes significant time, so sufficient parsing time must be reserved |
Segment Length | 800–1200 characters | Balances semantic completeness for both structured product parameters and unstructured event copy, to avoid content truncation that impacts retrieval |
Number of Retrieved Results | Top 8 | Covers retrieval needs for multiple categories of rural commercial bank marketing content, returning sufficient candidate matching results |
Similarity Threshold | 0.72–0.78 | Balances accurate matching and coverage of similar marketing content with different phrasing |
ENABLE_LOCAL_FILE_SYNC | Enabled | Adapts to the scenario where rural commercial bank marketing data is stored on intranet servers, supporting scheduled local file synchronization |
The parameter values provided on this page are general recommendations to serve as a starting point for configuration. Actual values are affected by material format, data volume, and business rules. Each scenario requires individual analysis. It is recommended to test on your own samples before finalizing settings.
Three Common Configuration Mistakes
- Issue: The FastGPT console cannot be accessed after opening port 3000 on the server and configuring
SERVER_ROOT_PASSWORD. Cause: Inbound rules for port 3000 were not added to the server security group for access. Modifying only the docker-compose.yml file will not take effect. - Issue: Knowledge base parsing progress stalls or produces no results after uploading PDF marketing materials. Cause: The local directory for the pdf-marker container was not correctly mounted in docker-compose.yml, leading to abnormal inter-container communication.
- Issue: Timeout errors are returned when calling a configured external large model channel. Cause: The API request timeout duration for the large model channel was not set to exceed the value of
PARSE_FILE_TIMEOUT_SECONDS, or no intranet proxy was configured to allow access to external APIs.
How to Confirm Configuration is Complete
- Run the
docker-compose pscommand. Confirm all FastGPT-related containers are in the Up state. - Upload a rural commercial bank-specific marketing PDF file. Review the parsing task status logs to confirm no timeout or format error messages are present.
- Submit a test retrieval request. Verify that the number of returned results matches the configured
Number of Retrieved Results. - Modify a single configuration parameter. Restart the corresponding container. Confirm the configuration takes effect and no error logs are generated in the console.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.