Combining Neural and Non-Neural Methods for Low-Resource Morphological Reinflection
Saeed Najafi, Bradley D. Hauer, Rashed Rubby Riyadh, Leyuan Yu, Grzegorz Kondrak · Proceedings of the · 2018
We describe our systems and results in the type-level low-resource setting of the CoNLL-SIGMORPHON 2018 Shared Task on Universal Morphological Reinflection.We test nonneural transduction models, as well as more recent neural methods.We also investigate the effect of leveraging unannotated corpora to improve the performance of selected methods.Our best system obtains the highest accuracy on 34 out of 103 languages.