University of Pittsburgh

Unsupervised morphology disambiguation using cross-lingual information

ISP graduate Student
Friday, April 22, 2016 - 12:50pm - 1:10pm

Computational analysis of morphology is the most fundamental step in Natural Language Processing (NLP). Current state of the art in corpus linguistic and NLP to deal with morphology disambiguation is the costly and time consuming process of manually preparing a very large annotated text corpus. Multi-lingual and cross-lingual NLP as a way to induce implicit features of a language from a comparable text in another language can remarkably reduce the manual works needed in traditional NLP. In this paper, we focus on morphology disambiguation in Persian and propose a cross-lingual approach to automatically prepare a morphologically disambiguated corpus from a comparable bi-lingual corpus.

Copyright 2009–2020 | Send feedback about this site