SPAACy – A semi-automated tool for annotating dialogue acts

This article reports on a pilot project which aims at creating a speech-act annotated training corpus for service dialogue systems. In order to achieve this aim, an annotation tool, which allows us to automate large parts of the annotation, is being developed. This tool converts text-based transcriptions into XML and applies different levels of markup to each dialogue, so that there remains as little post-editing to be done as possible. The project also aims at developing a relatively generic mark-up scheme that may be applied to different domains without needing a large degree of adaptation. This article describes aspects of the grammar ‘controlling/governing’ the tool and how this grammar ‘interacts’ with the general strategies employed in the annotation.