Introduction

This release focuses on strengthening database integration capabilities and revamping the AI search infrastructure. We have introduced the ability to convert existing database tables directly into datasets and constructed an OpenSearch environment as a foundation for improving AI retrieval accuracy. Additionally, we have refined the cross-system identity synchronization for administrators and optimized the access flow for non-logged-in users.

Release Date: 2026.03.11

Modules Version: - Stellar: V1.2.0

• Data Planet: V2.2.0
• MCP Server: V0.0.3
• Galaxy: V0.1.1 (Unchanged)

New Features

• Establish OpenSearch High-Level Retrieval Environment: Officially deployed the OpenSearch environment and integrated the latest Embedding models, supporting vector retrieval and semantic search. This serves as a foundation for enhancing AI's understanding and searchability of datasets in future updates.

Improvements & Enhancements

• Strengthened Admin AI Settings Interface: A new AI service settings screen has been added to the management system, allowing administrators to intuitively adjust AI model configuration parameters without technical commands.
• Enhanced Cross-System Connection Performance and Stability: Optimized the cross-system connection mechanism to ensure that user connection status is maintained seamlessly when switching between different functional web pages, reducing the need for repeated logins.
• Automated OpenAI API Key Management Tool: Developed a specialized utility to assist in the rapid update of API keys required by the AI inference engine, simplifying maintenance costs and shortening downtime.

Bug Fixes

• Fixed Guest Access Flow: Corrected the issue where non-logged-in users were incorrectly redirected to the login page when accessing public datasets; anonymous users can now browse public data normally.
• Resolved Local Cross-Data Exploration Permission Anomalies: Fixed a bug where reading permission errors occurred during cross-data exploration of specific datasets within one's own Planet under certain conditions.
• Fixed Exploration Compatibility After Database Replacement: Optimized data Join behavior following database architecture adjustments to ensure data is not lost due to type differences during complex data linking.
• Resolved SDMX OECD Cache Creation Failures: Fixed cache conflicts caused by duplicate IDs in external source data; OECD data flows can now complete cache creation and import more stably.