Probabilistic latent semantic analysis as a potential method for integrating spatial data concepts

Abstract

In this paper we explore the use of Probabilistic Latent Semantic Analysis (PLSA) as a method for quantifying semantic differences between land cover classes. The results are promising, revealing ‘hidden’ or not easily discernible data concepts. PLSA provides a ‘bottom up’ approach to interoperability problems for users in the face of ‘top down’ solutions provided by formal ontologies. We note the potential for a meta-problem of how to interpret the concepts and the need for further research to reconcile the top-down and bottom-up approaches.