The Author Perspective Model for Classifying Deontic Modality in Events
David B. Bracewell, David Hinote, Sean Monahan · The Florida AI Research Society · 2014
In this paper we present a generative model entitled the Author Perspective Model for the classification of deontic modality in event mentions. In the model modals, adverbials, and predicates associated with an event mention are generated by either a topic or author perspective where the author perspective is one of the three high level categories of deontic modality. We train the model with data gathered by a small set of seed phrases for each of the deontic modality categories. Our results show that we are able to classify the category of deontic modality with a micro-averaged F-Measure of 67.3%.