Bilateral Spatial Reasoning about Street Networks: Graph-based RAG with Qualitative Spatial Representations

Bilateral Spatial Reasoning about Street Networks: Graph-based RAG with Qualitative Spatial Representations R. Moratz, N. Daute, J. Ondieki, M. Kattenbeck, M. Krajina, I. Giannopoulos December 18, 2025 Abstract This paper deals with improving the capabilities of Large Language Models (LLM) to provide route instructions for pedestrian wayfinders by means of qualitative spatial relations. We use a method called Retrieval- Augmented Generation (RAG). RAG supports the LLM with context in- formation based on the specific query. We assess the impact the added information has on model performance for generating pedestrian route instructions. Our findings encourage further integration of qualitative spatial data into LLM applications—potentially benefiting areas such as digital navigation aids, smart city tools, and accessibility technologies. 1 Introduction Up until now, Large Language Models show rather weak performance in pro- viding route instructions to pedestrian wayfinders. For example we can ask an LLM (ChatGPT-4o by OpenAI) for the direction from M¨unster central station to nearby street Hafenweg (see figure1). The model responded quickly with a seemingly convincing set of directions for the given problem. However, upon closer inspection, it becomes evident that LLM navigation is heavily prone to hallucinations and other errors. In the first step, the model suggests leaving the central station onto ”Willy-Brandt-Allee”, which is a street that does not exist in M¨unster. Subsequently, the model sug- gests walking toward the city center, while the Hafenweg is actually located away from the city center. Next, turning onto the ”Albersloher Weg” is sug- gested. While this road exists in M¨unster and is close to the actual destination, it does not make sense to take the road for the given navigation task. The same can be said for the following two steps, where turning onto ”Aaseepark” and ”Hammer Straße” is suggested, with neither of them being sensible choices for the problem, while also being completely disconnected, making navigation along the suggested route impossible. Finally, the model suggests turning onto 1 arXiv:2512.15388v1 [cs.AI] 17 Dec 2025 Figure 1: Route instruction generated by LLM the destination street ”Hafenweg”, which is the correct destination, but impos- sible given the previous incorrect steps. When comparing this route proposed by ChatGPT to a route provided by Google Maps, the problem becomes even more evident. In the figure, the correct route from ”Start” to ”Finish” is shown in green: A few simple turns from the Central Stations is all it takes to reach the destination. The generated LLM route displayed in red on the map is how- ever far from correct: Mutltiple disconnected segments show a complete lack of understanding the actual navigation task, with only one of them being close to the actual route. Observations like these highlight the current limitations of LLMs in navigation performance, and motivated us to find approaches to boost their capabilities in this domain. 2 Figure 2: Map detail showing erroneous route instruction 3 This paper deals with improving the capabilities of Large Language Models to provide route instructions for pedestrian wayfinders by means of qualita- tive spatial relations. We use a method called Retrieval-Augmented Generation (RAG). RAG supports the LLM with context information based on the spe- cific query. For this purpose we use a qualitative spatial representation frame- work which is based on oriented line segments (dipoles) as basic entities [4]. In our context a qualitative representation provides mechanisms which characterize central essential properties of objects or configurations. A quantitative repre- sentation in contrast establishes a measure in relation to a unit of measurement which has to be generally available. Qualitative spatial spatial representations usually deal with elementary objects (e.g., positions, directions, regions) and qualitative relations between them (e.g., ”adjacent”, ”on the left of”, ”included in”). 2 Graph-Based Retrieval Augmented Genera- tion (Graph-RAG) Graph-based Retrieval Augmented Generation represents an advanced evolu- tion of traditional RAG systems that leverages knowledge graphs to enhance information retrieval and generation quality [1]. With Graph-RAG language model receives not just text chunks, but structured information including en- tity mentions and their properties, explicit relationships between entities, graph paths showing logical connections, and multi-level abstractions (detailed facts and high-level summaries). Traditional RAG retrieves relevant text chunks from a vector database us- ing semantic similarity, then feeds these chunks to a language model for gen- eration. Graph-RAG enhances this by organizing information in a knowledge graph structure, where entities are nodes and relationships are edges, enabling more sophisticated retrieval strategies. The system first constructs a knowledg

This content is AI-processed based on open access ArXiv data.