The last few decades have seen impressive improvements in several areas of Natural Language Processing. Nevertheless, getting a computer to make sense of the discourse of utterances in a text remains challenging. Several different theories which aim to describe and analyze the coherent structure of a well-written text exist, but with varying degrees of applicability and feasibility for practical use.
This book is about shallow discourse parsing, following the paradigm of the Penn Discourse TreeBank, a corpus containing over 1 million words annotated for discourse relations. When it comes to discourse processing, any language other than English must be considered a low-resource language. This book relates to discourse parsing for German. The limited availability of annotated data for German means that the potential of modern, deep-learning-based methods relying on such data is also limited. This book explores to what extent machine-learning and more recent deep-learning-based methods can be combined with traditional, linguistic feature engineering to improve performance for the discourse parsing task.
The end-to-end shallow discourse parser for German developed for the purpose of this book is open-source and available online. Work has also been carried out on several connective lexicons in different languages. Strategies are discussed for creating or further developing such lexicons for a given language, as are suggestions on how to further increase their usefulness for shallow discourse parsing.
The book will be of interest to all whose work involves Natural Language Processing, particularly in languages other than English.