|Antoine Venant, Nicholas Asher, Philippe Muller, Pascal Denis, Stergos Afantenos|
Several discourse annotated corpora now exist for NLP. But they use different, not easily comparable annotation schemes: are the structures these schemes describe incompatible, incomparable, or do they share interpretations? In this paper, we relate three types of discourse annotation used in corpora or discourse parsing: (i) RST, (ii) SDRT, and (iii) dependency tree structures. We offer a common language in which their structures can be defined and furnished a range of interpretations. We define translations between RST and DT preserving these interpretations, and introduce a similarity measure for discourse representations in these frameworks. This will enable researchers to exploit different types of discourse annotated data for automated tasks.