SCIS Graduate Students and Faculty Have Impressive Showing at COSIT
From October 12-16, the School of Information and Computing Science at the University of Maine sent several researchers to COSIT, one of the most significant conferences in the world for spatial information theory. Established in 1993 and only held every two years, COSIT seeks to bring together scientists from different disciplines for an intensive and deep exchange of knowledge and research. The conference is a single-track meeting that focuses on recent, innovative and significant contributions in the field. This year, University of Maine researchers including professors, graduate students and undergraduate students, presented an impressive three of the total 24 papers at the conference.
The first paper presented by University of Maine researchers is titled “From Metric to Topology: Determining Relation in Discrete Spaces.” This paper details a set of metrics that can uniquely identify simple topological relations between pixelized objects where the boundary is considered to be the outer set of pixels. This research is particularly pertinent in applications where images need to be analyzed for natural language interfaces, such as with drones, robots, satellites, and other camera-assisted technologies. This work can also have significant application with forest fire management through imagery and also in the understanding of temporal patterns in meteorology and other areas. The method proposed identifies the relation between the objects, which comes from common vocabulary within whichever language the user would prefer the information in. This publication is the culmination of a project with Upward Bound Math-Science student Jordan Barrett, now attending Syracuse University for Mathematics and Physics. Max Egenhofer also collaborated on this paper.
The second paper presented is titled “Swiss Canton Regions: A Model for Complex Objects in Geographic Partitions.” This paper constructs a method for maintaining topological equivalence up to the level of homeomorphism. In other words, this method helps to determine whether two areal objects separate a space in exactly the same way conceptually. This work is the culmination of a group research course in Spatial Reasoning, one of the core benefits of attending the Spatial Informatics program at the University of Maine. The benefit of this research is in the ability to construct queries that go beyond the simple topological queries in commercial GIS and enter into a more complex representation where separations, holes, and self-adjacency matter. Areas such as land cover analysis can gain substantially from this sort of approach. Furthermore, these types of queries are essential in cover analysis for sensor networks, helping to identify coverage issues in the deployment. Professor Max Egenhofer and current Ph.D. student Joshua Lewis, M.S. student Shirly Stephen, and former student Mark Plummer are also authors on the paper.
The third paper presented at COSIT is titled “Shape Similarity Based on the Qualitative Spatial Reasoning Calculus eOPRA.” The paper investigated the use of qualitative spatial representations (QSR) about relative direction and distance for shape representation. The new approach has the advantage that it can generate prototypical shapes from the authors’ abstract representation in first-order predicate calculus. Using the conceptual neighborhood which is an established concept in QSR the authors can directly establish a conceptual neighborhood between shapes that translates into a similarity metric for shapes. This similarity measure was applied to a challenging computer vision problem and achieved promising first results. This paper was authored by Ph.D. student Christopher Dorr, Professor Reinhard Moratz, and Longin Jan Latecki of the Temple University.
The fourth paper is entitled “What is in a Contour Map? A Region-based Logical Formalization of Contour Semantics.” This paper analyses and formalizes contour semantics in a first-order logic ontology that forms the basis for enabling computational common sense reasoning about contour information. The elicited contour semantics comprises four key concepts – contour regions, contour lines, contour values, and contour sets – and their subclasses and associated relations, which are grounded in an existing qualitative spatial ontology. All concepts and relations are illustrated and motivated by physical-geographic features identifiable on topographic contour maps. The encoding of the semantics of contour concepts in first-order logic and a derived conceptual model as basis for an OWL ontology lay the foundation for fully automated, semantically-aware qualitative and quantitative reasoning about contours. This paper was authored by Professor Torsten Hahmman and Lynn Usery of the U.S. Geological Survey.
Professor Max Egenhofer of the School of Computing and Information Science remarked at the impressive nature of not only the quality of the research produced by UMaine researchers, but also at how influential the institution has become in the field of spatial information theory. “The Conference on Spatial Information Theory is the intellectual highlight every two years in the field of Spatial Information Science. The papers presented there are of journal-like quality as they go through a rigors reviewing and selection process. The fact that this year four of the twenty-two papers that were chosen for publication and presentations were authored by UMaine researchers indicates most strongly what a powerhouse UMaine is this field. I was particularly delighted that three of these fours papers were authored or co-authored by a total of five of our graduate students, plus another co-author who is currently an undergraduate student. Many of the conference attendees approached me, commenting on the excellent presentations that our students gave. This involvement of students into top-level research highlights the kind of education that students get in our environment.”
Information on the conference and the conference program may be found at http://www.unm.edu/~sfreunds/COSIT_main/Home.html.