I am responding to a paper by Sieber et al. (1) published in PNAS. Although spatial ontology and semantic research are not new, a fundamental scientific question has been ignored for a long time.
According to Gruber (2), ontology is a “body of formally represented knowledge [that] is based on a conceptualization: the objects, concepts, and other entities that are assumed to exist in some area of interest and the relationships that hold among them” (Genesereth and Nilsson, 1987; ref. 2). Although “ontologies are also not limited to conservative definitions, that is, definitions in the traditional logic sense that only introduce terminology and do not add any knowledge about the world” (Enderton, 1972; ref. 2), eventually in computer science and artificial intelligence (AI) community, the so-called relationships only target the logical relationships between the concepts.
Indeed, besides the logic sense, the spatial relationships between spatial features are the idiosyncratic identity of GIScience. Spatial relationships normally include such relationships of equal, within, contain, touch, disjoin, intersect, union, exclusive, and difference between spatial features. The scientific foundation to determine spatial relationships is computational geometry; that is to say, we cannot use logics to derive and determine the spatial relationships.
If spatial ontology has nothing about the spatial relationships but only deals with the conceptual matchmaking through logics, then the disciplinary identity of spatial science is missing, although conceptual relationships are important. This problem is common and not only in the paper by Sieber et al. (1). Previous works (3–5) on geospatial ontology may have been problematic as well; they all dealt with the logical relationships between spatial concepts, but the spatial relationships between spatial objects were ignored. Obviously, researchers forgot that ontologies are not just limited to the logic sense.
If a spatial ontology does not cover spatial relationships but is full of logical relationships, it cannot be called a spatial ontology; instead, it looks like a conceptual and logic game. Although existing semantic technology may be able to handle logic rules, unfortunately, it cannot support reasoning spatial relationships, which are based on computational geometry, a different scientific domain from logics. To develop spatial ontology, we need new technologies in which the key is to determine and reason the spatial relationship by computational geometry and not just conceptual and logical relationships.
Footnotes
The author declares no conflict of interest.
References
- 1.Sieber RE, Wellen CC, Jin Y. Spatial cyberinfrastructures, ontologies, and the humanities. Proc Natl Acad Sci USA. 2011;108:5504–5509. doi: 10.1073/pnas.0911052108. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Gruber T. 1992. What is an Ontology? Available at http://www-ksl.stanford.edu/kst/what-is-an-ontology.html. Accessed June 29, 2011.
- 3.Fonseca F, Egenhofer M. Ontology-driven geographic information systems. In: Medeiros CB, editor. Proceedings of the Seventh Association for Computing Machinery (ACM) Symposium on Advances in Geographic Information Systems. New York, NY: ACM; 1999. pp. 14–19. [Google Scholar]
- 4.W3C 2007. W3C Geospatial Ontologies. Available at http://www.w3.org/2005/Incubator/geo/XGR-geo-ont/. Accessed June 29, 2011.
- 5.National Science Foundation Award #0955816. Available at http://www.nsf.gov/awardsearch/showAward.do?AwardNumber=0955816&WT.z_pims_id=502112. Accessed June 29, 2001.