dc.description.abstract |
Machine translation is an application of NLP, which can be used to translate text from one
natural language to another natural language. This research attempted to design Amharic-Kistanigna bidirectional machine translation. Previously, there is no research conducted on
machine translation between Kistanigna and Amharic. Due to an increase in the number of
language users, to addressing the issues of the endangered of the Kistanigna language and to
increases the content of the language in web, it is essential to develop machine translation
between Kistanigna and Amharic languages. Different official documents, news and other
written texts in both languages needed to be translated in order to share information. Neural
Machine Translation (NMT) is a recently proposed approach to machine translation (MT) that
has achieved the state-of-the-art translation quality in recent years. Unlike traditional MT
approaches, NMT aims to create a single neural network that can be tuned collaboratively to
maximize translation performance. So, the aim of this study is to develop Amharic-Kistanigna
bi-directional machine translation using Deep learning.
We conduct the experiments using five encoder decoder models. These are LSTM, Bi-LSTM,
LSTM+attention, CNN+attention, and Transformers. To conduct our experiments, we have used
9,225 parallel sentences using both word and morpheme translation unit. In order to segment our
morpheme data, we have used morfessor tool. To propose an optimal model with best translation
unit we consider efficiency (training time, memory usage, and BLEU score). Finally, we have
proposed morpheme-based bi-directional machine translation using Transformers model with a
BLEU score of 21.31 and 22.40 from Amharic-Kistanigna and Kistanigna-Amharic translation
respectively. The major weakness of the study is unavailablity of enough dataset to conduct an
extensive experiment. As a result, there is a need to prepare parallel corpora for conducting
similar research.
Keywords: Amharic-Kistanigna Machine Translation, Bi-Directioanl Machine Translation,
Neural Machine Translation, Deep Learning Approach, LSTM, Bi-LSTM, LSTM+attention,
CNN+attention, Transformers. |
en_US |