On the Formulation of Conceptual Spaces for Land Cover Classification Systems

Cognitive approaches to knowledge representation improve man-machine communication, as they are close to human reasoning. Conceptual spaces have been proposed as one such knowledge formalization method. Our research investigates the theory of conceptual spaces as a methodology for implementing semantic reference systems. Conceptual spaces are spanned by quality dimensions. Concepts are represented as regions in n-dimensional spaces and instances as n-dimensional vectors. The land cover domain is chosen for applying this theory with the view to formulating a conceptual space from textual descriptions. Based on a land cover classification system and its descriptions, the methodology for extracting the quality dimensions is demonstrated and their measurement scales are discussed. The usefulness of formalizing a classification system as conceptual space is demonstrated in the process of semantically transforming instances from one classification system to another.