Cette dernière séance est consacrée aux approches informatiques de l'analyse de l'argumentation, et plus spécifiquement aux modèles calculatoires de l'argumentation et d'extraction automatique des schémas argumentatifs à partir des textes en langue naturelle.
Intervenants:
1) Srdjan Vesic, CNRS, CRIL, France
Introduction to Computational Models of Argument
Abstract:
In this talk we present an overview of computational approaches to argumentation. The most common model used to formalise argumentation by computer scientists is a directed graph. The nodes of this graph represent arguments and the edges attacks between them. It is also possible to add another relation between arguments, called support. A well-studied question in the literature is: given a graph representing arguments and attacks between them, how to calculate the set(s) of acceptable arguments? The talk is devoted to presenting the most influential approaches that aim at answering this question.
2) Marie-Francine Moens, Katholieke Universiteit Leuven, Belgium
Argumentation Mining: What is Next?
Abstract:
In this lecture we give a definition of argumentation mining from text and give an overview of the state-of-the-art in this emerging field of natural language processing. We focus on two important machine learning problems and on possible solutions. One problem regards the recognition of argumentation structures and the recognition of the relationships between argumentation components. The second regards learning of the right representations of words, phrases and other textual units that capture the necessary world and common sense knowledge, and that facilitate the recognition of the relationships.
3) Iryna Gurevych, Technische Universitaet Darmstadt, Germany
Structure Identification and Quality Assessment of Arguments in Argumentative Essays
Abstract:
In this talk, we provide an overview of the methods developed in the context of Argumentative Writing Support (AWS) at the UKP Lab, Technische Universität Darmstadt. Our research employs a corpus of argumentative essays. We present the results of an annotation study on argumentation structures, introduce an argumentation structure annotated corpus, and present a novel end-to-end argumentation structure parser for extracting micro-level argumentation structures. In addition, we introduce two novel tasks and our experimental results on quality assessment of natural language arguments: identifying myside bias in argumentative essays, and identifying insufficiently supported arguments. In conclusion, we outline several related research efforts on argumentation analysis at the UKP Lab involving further user groups and comment on the emergence of an international research community of Computational Argumentation.
URI/Permalink: