Exploiting metadata for ontology-based visual exploration of weakly structured text documents

Abstract

A large amount of strategically relevant business information is contained in unstructured texts. While information brokering approaches are used to contextualize such documents and to generate metadata, text mining is used to explore large document spaces. So far, little attention has been paid on a value-adding combination of these technologies. We show how metadata and documents can be complementarily represented and used interactively to support users in text corpus analysis. We present a text analysis portal which displays inter-document similarity by means of so-called document maps, complemented by a display of the domain ontology and metadata-based access methods.