Intelligent Information Management

Bavarian Environmental Atlas is the official Environmental Platform Geospatial Platform for the Free State of Bavaria. It provides public access to spatial datasets covering flood zones,nature reserves,water protection,and soil types as a form of multiple layers. These Dataset is critical for urban planning, environmental monitoring, and regulatory decision making.Recent models in Larger Language Models (LLM) and Retrieval Augmented Generation (RAG) have demonstrated significant potential for enabling natural language interface for complex,domain specific information systems.This external master’s thesis,conducted in collaboration with con terra for LFU Bayern,investigates the practical integration of these technologies into live production geospatial platforms. Bavarian Environmental Atlas already has the public relevance for understanding geospatial information, but the effective use of the platform requires GIS expertise. Users must navigate complex hierarchical menus, know the technical names of map layers, and understand spatial query concepts such as buffering and intersection. This sometimes creates a significant barrier to non-expert users including policy makers, urban planners,citizens who are the primary audience for publicly mandated environmental data. This thesis aims to design, implement, and evaluate RAG enhanced LLM based natural language interface integrated directly with Umweltatlas enabling non expert users to control the map through plain queries.