Automatic Detection of Modality with ITGETARUNS

In this paper we present a system for modality detection which is then used for Subjectivity and Factuality evaluation. The system has been tested lately on a task for Subjectivity and Irony detection in Italian tweets, where the performance was 10 and 4, respectively, over 27 participants overall. We will focus our paper on an internal evaluation where we considered three national newspapers Il Corriere, Repubblica, Libero. This task was prompted by a project on the evaluation of press stylistic features in political discourse. The project used newspaper articles from the same sources over a period of three months, thus including latest political 2013 governmental crisis. We intended to produce a similar experiment and evaluate results in comparison with previous 2011 crisis. In this evaluation, we focused on Subjectivity, Polarity and Factuality which include Modality evaluation. Final graphs at the end of the paper will show results confirming our previous findings about differences in style, with Il Corriere emerging as the most atypical.