Ontology-driven data mining platform for fuzzy classification of mental maps

Abstract

Mental maps are valuable material for Digital Humanities research since they represent a summary of humans spatial experience reflected in their minds. Contributing to this research, we developed a high-level Web-based software platform that allows to collect drawings of mental maps and to perform corresponding data mining and fuzzy classification. The novelty of the proposed platform is the ontology-based integration of mental maps drawing engine and data mining engine, wherein all the essential steps of data mining, including data acquisition, transformation, fuzzy classification, and visual analytics are driven by ontologies. The platform consists of a high-level graphical editor to draw maps and a data flow diagram editor to define the data mining pipeline. The operators available to construct this pipeline are described by ontology, which ensures the platform's extensibility on the knowledge base level. Thereby, the platform created can be used not only for Digital Humanities research but also for testing and evaluation of new data mining and fuzzy classification methods. Currently, we have evaluated weighted fuzzy pattern matching for mental maps fuzzy classification and confirmed the reasonable performance of this method.