NSF research grant to advance spatial artificial intelligence
“Just over the hill” or “downstream” are phrases commonly used to communicate location. But how can these basic instructions be used for computing?
With a two-year, $175,000 research grant from the National Science Foundation, Torsten Hahmann, an assistant professor in the School of Computing and Information Science, will explore the implications of informal spatial descriptions on computing and artificial intelligence.
The project, “Empowering Multi-Conceptual Spatial Reasoning with a Repository of Qualitative and Quantitative Spatial Ontologies,” will develop techniques for computers to flexibly and reliably deal with a wide range of informal spatial descriptions, such as “east of the road” or “at the south shore of the lake.”
Hahmann will investigate computational mechanisms that connect such descriptions to more traditional coordinate-based spatial information used in satellite mapping or GPS devices.
His research will also develop methods that enable computers to figure out what implicit assumptions are made in such informal spatial descriptions. For example, a simple piece of information such as “X is contained in the lake” may take on very different spatial meanings in different contexts. If X is a bay, it means that it consists of a portion of the lake’s water, whereas if X is an island, it means it is surrounded by the lake water, but does not consist of water itself.
Once a computer better understands such descriptions, it can add missing information from existing maps. In the long term, the research will make computational tools for recording, processing, and searching through spatial information much more powerful and user-friendly.
The project will support two graduate students and offer opportunities for undergraduate students to get involved in cutting-edge research.
More information about the project is online.