OntoSDM: an approach to improve quality on spatial data mining algorithms

Abstract

The increase in new electronic devices had generated a considerable increase in obtaining spatial data information; hence these data are becoming more and more widely used. As well as for conventional data, spatial data need to be analyzed so interesting information can be retrieved from them. Therefore, data clustering techniques can be used to extract clusters of a set of spatial data. However, current approaches do not consider the implicit semantics that exist between a region and an object’s attributes. This paper presents an approach that enhances spatial data mining process, so they can use the semantic that exists within a region. A framework was developed, OntoSDM, which enables spatial data mining algorithms to communicate with ontologies in order to enhance the algorithm’s result. The experiments demonstrated a semantically improved result, generating more interesting clusters, therefore reducing manual analysis work of an expert.

Este artigo foi publicado em Lecture Notes in Computer Science, a partir da apresentação do mesmo na 41st International Conference on Current Trends in Theory and Practice of Computer Science, Pec pod Sně kou, Czech Republic, January 24-29, 2015. Proceedings