Multilingual Extension of Dependency Parsing and Annotation

Abstract

Dependency parsing is considered as the state of the art technology for a better information extraction methodology in Natural Language Processing. With the ever-growing need for linguistic analysis for different languages, the demand for multilingual dependency parsing has increased dramatically. In this research work we studied a novel collection of treebanks with homogeneous syntactic dependency annotation [1] for six languages along with other recent techniques in this area. We investigated the possibility of adding new languages in this module and successfully added universal Bangla dependency annotation. Additionally, we combined simple and complex feature representations to improve parsing output.