Analysing Discourse Automatically with Multiple Objectives – AnDiAMO

Documents are not random sequences of sentences, but rather spans of texts are linked together into a discourse structure to make coherent and meaningful documents. Developing systems able to interpret documents and to extract structured information is a challenge in natural language processing that requires high-performing discourse parsers. However, performance are still low and current work does not make it clear where the problem lies. Moreover, studies focus on English monologues from news, which limits their use and their robustness. Finally, empirical studies rarely try to inform theoretical models for discourse. We propose a general framework here called multi-objective learning to tackle these issues. We will build systems seeking to achieve several goals as a way to provide robust systems for multiple languages, domains and modalities, to investigate data representation and problem modelling, to improve evaluation, and to shed some light on the theoretical divergences.