Assessment of semantic similarity between land use/land cover classification systems

Abstract

One difficulty in sharing datasets between research projects and research teams is variation in the semantic content of the data expressed as differences in categorization. One step towards achieving semantic interoperability is to have means to measure the degree of semantic similarity between existing datasets and the target categorization of the new study. A variety of environmental and urban analyses use land use/land cover data to describe the natural environment. This paper focuses on measuring semantic similarity between categories in different land use/land cover classification systems. A modified feature-based approach is employed. Semantic components of a category are weighted to emphasize critical features in the categories definition. A case study is presented to demonstrate how semantic similarity is measured between target categories in a specific study and categories already in use in existing land use/land cover classification systems.