Application of fuzzy sets in data-to-text systems

Abstract

This PhD dissertation addresses the convergence of two distinct paradigms: fuzzy sets and natural language generation. The object of study is the integration of fuzzy set-derived techniques that model imprecision and uncertainty in human language into systems that generate textual information from numeric data, commonly known as data-to-text systems. This dissertation covers an extensive state of the art review, potential convergence points, two real data-to-text applications that integrate fuzzy sets (in the meteorology and learning analytics domains), and a model that encompasses the most relevant elements in the linguistic description of data discipline and provides a framework for building and integrating fuzzy set-based approaches into natural language generation/data-to-ext systems.