Speaker: Djamé Seddah, maître de conférences at Université Paris-Sorbonne, Paris, France

Abstract: Parsing predicate-argument structures in a deep syntax framework requires graphs to be predicted. Argument structures represent a higher level of abstraction than the syntactic ones and are thus more difficult to predict even for highly accurate parsing models on surfacic syntax. In this talk I investigate deep syntax parsing, using both a French data set (Ribeyre et al., 2014a) and an English data set (Oepen et al, 2015). We demonstrate that the use of topologically different types of syntactic features, such as dependencies, tree fragments, spines or syntactic paths, brings a much needed context to the parser. Our higher-order parsing model, gaining thus up to 4 points on one language, establishes the state of the art for parsing such deep syntactic structures. I will also describe the process of creating a deep syntax graph-based treebank for French. In addition to this, I will also present our results and models at the recent Extrinsic Parsing Shared Task (EPE 17) where combined systems from Stanford and our team were tested on a various set of graph-based data sets and established a new state-of-the-art. (joint work with Eric de la Clergerie, Corentin Ribeyre, Benoit Sagot, Marie Candito, Sebastian Schuster)