Reasoning over higher-order qualitative spatial relations via spatially explicit neural networks

Abstract

Qualitative spatial reasoning has been a core research topic in GIScience and AI for decades. It has been adopted in a wide range of applications such as wayfinding, question answering, and robotics. Most developed spatial inference engines use symbolic representation and reasoning, which focuses on small and densely connected data sets, and struggles to deal with noise and vagueness. However, with more sensors becoming available, reasoning over spatial relations on large-scale and noisy geospatial data sets requires more robust alternatives. This paper, therefore, proposes a subsymbolic approach using neural networks to facilitate qualitative spatial reasoning. More specifically, we focus on higher-order spatial relations as those have been largely ignored due to the binary nature of the underlying representations, e.g. knowledge graphs. We specifically explore the use of neural networks to reason over ternary projective relations such as between. We consider multiple types of spatial constraint, including higher-order relatedness and the conceptual neighborhood of ternary projective relations to make the proposed model spatially explicit. We introduce evaluating results demonstrating that the proposed spatially explicit method substantially outperforms the existing baseline by about 20%.

Keywords:

• Geospatial knowledge graphs
• higher-order spatial interactions
• qualitative spatial representation and reasoning
• spatially explicit methods
• GeoAI

Data and codes availability statement

The data and codes that support the findings of this study are available with the identifier(s) at the link https://doi.org/10.6084/m9.figshare.13350737.v2.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 https://www.w3.org/RDF/.

2 Note that the straight-line based segmentation in Figure 1 is replaced by a path-based segmentation in order to consider the complexity of real world data when generating relation statements for experiments in this paper. More details will be discussed in Section 5.1.

3 Strictly speaking, the figure includes line segments but as these also have a spatial extent in geographic space we will consider them as regions here.

4 https://wiki.dbpedia.org/projects/dbpedia-places.

5 https://www.wikipedia.org/.

6 https://developers.google.com/maps/documentation/directions/overview.

7 https://docs.ray.io/en/latest/tune/index.html.

Additional information

Funding

This work is funded by the National Science Foundation’s Convergence Accelerator Program under [Grant No. 1936677 and No. 2033521].

Notes on contributors

Rui Zhu

Rui Zhu is a Postdoctoral Scholar at the Center for Spatial Studies, University of California, Santa Barbara. He obtained a PhD in Geography from the University of California, Santa Barbara. He also holds a master degree in Information Sciences from the University of Pittsburgh and a bachelor degree in Information Management and Information Systems from Shanxi University of Fiance and Economics. Rui has expertise in spatial statistics, geospatial semantics, knowledge graphs, and GeoAI. His work has been applied to urban planning, global health, environmental intelligence, as well as humanitarian relief.

Krzysztof Janowicz

Krzysztof Janowicz is a Professor for Geoinformatics at the University of California, Santa Barbara and director of the Center for Spatial Studies. He is also a Professor at the Department of Geography and Regional Research, University of Vienna. His research focuses on how humans conceptualize the space around them based on their behavior, focusing particularly on regional and cultural differences with the ultimate goal of assisting machines to better understand the information needs of an increasingly diverse user base. Janowicz’s expertise is in knowledge representation and reasoning as they apply to spatial and geographic data, e.g. in the form of knowledge graphs.

Ling Cai

Ling Cai is a PhD Candidate at the Space and Time Knowledge Organization Lab, Department of Geography, University of California, Santa Barbara. She obtained her M.S. degree in Geographical Information Science from Chinese Academy of Sciences and B.S. degree from Wuhan University. Her research interests include qualitative spatial temporal reasoning, temporal knowledge graph, neuro-symbolic AI, and urban computing.

Gengchen Mai

Gengchen Mai is a Postdoctoral Scholar at the Stanford AI Lab, Department of Computer Science, Stanford University. He is also affiliated with the Stanford Sustainability and AI Lab. He obtained his PhD degree in Geography from the Department of Geography, University of California, Santa Barbara in 2021 and B.S. degree in GIS from Wuhan University in 2015. His research mainly focuses on spatially-explicit machine learning, geospatial knowledge graph, and geographic question answering.