Translating Low-Resource Languages by Vocabulary Adaptation from Close Counterparts

Abstract

Some natural languages belong to the same family or share similar syntactic and/or semantic regularities. This property persuades researchers to share computational models across languages and benefit from high-quality models to boost existing low-performance counterparts. In this article, we follow a similar idea, whereby we develop statistical and neural machine translation (MT) engines that are trained on one language pair but are used to translate another language. First we train a reliable model for a highresource language, and then we exploit cross-lingual similarities and adapt the model to work for a close language with almost zero resources. We chose Turkish (Tr) and Azeri or Azerbaijani (Az) as the proposed pair in our experiments. Azeri suffers from lack of resources as there is almost no bilingual corpus for this language. Via our techniques, we are able to train an engine for the Az→English (En) direction, which is able to outperform all other existing models.

A Template Based Approach for Training NMT for Low-Resource Uralic Languages - A Pilot with Finnish

“…This means learning some of its characteristics (e.g. syntax and morphology), and transferring them to the low-resource MT model [14,20,21,26]. These approaches do not necessarily use parallel texts of the low-resource language as they rely on the resource-rich language, which could result in a poor coverage of the morphology of the low-resource language.…”

A Multitask-Based Neural Machine Translation Model with Part-of-Speech Tags Integration for Arabic Dialects

“…In the encoder-decoder architecture which was discussed by Peyman [24], two recurrent neural networks (RNNs) are trained together to maximize the conditional probability of a target sequence (candidate translation) y = y 1 , ... y m , given a source sentence x = x 1 , ... x n . Input words are sequentially processed consecutively until the end of the input string is reached.…”

A Neural Machine Translation Model for Arabic Dialects That Utilizes Multitask Learning (MTL)

“…In the encoder-decoder architecture which was discussed by Peyman et al [21], two recurrent neural networks (RNNs) are trained together to maximize the conditional probability of a target sequence (candidate translation) y = y 1 ,…, y m , given a source sentence x = x 1 ,…, x n . Input words are sequentially processed consecutively until the end of the input string is reached.…”