Seamless Tree Binarization for Interactive Predictive Parsing

Ricardo Sánchez-Sáez, Luis A. Leiva, Joan-Andreu Sánchez, José-Miguel Benedí

Abstract: This paper introduces a seamless method for tree binarization/debinarization that is employed within the Interactive Predictive Parsing framework for tree annotation. This novel method allows that, while the human annotator verifies and corrects standard non-binary trees, the parse engine can work with parsing algorithms that process and produce binary trees, such as a CYK-Viterbi based parser. Within the Interactive Predictive Parsing framework the user is tightly integrated into the interactive parsing system, in contrast with the traditional post-editing approach. User feedback for tree correction and validation is provided by means of natural mouse gestures and keyboard strokes.

Index Terms: parsing, interactive predictive parsing, syntactic tree annotation, tree binarization.

Full Paper