Constructing text sense representations
Ronald Winnemöller · 2004
In this paper we present a novel approach to map textual entities such as words, phrases, sentences, paragraphs or arbitrary text fragments onto artificial structures which we call "Text Sense Representation Trees" (TSR trees). These TSR trees represent an abstract notion of the meaning of the respective text, subjective to an abstract "common" understanding within the World Wide Web. TSR Trees can be used to support text and language processing systems such as text categorizers, classifiers, automatic summarizers and applications of the Semantic Web. We will explain how to construct the TSR tree structures and how to use them properly; furthermore we describe some preliminary evaluation results.