Selecting Parallel In-domain Sentences for Neural Machine Translation Using Monolingual Texts

TL;DR

This paper presents a method to select in-domain parallel sentences from generic corpora using cosine similarity with monolingual domain data, improving domain-specific machine translation performance.

Contribution

It introduces a simple, effective sentence ranking method for selecting high-quality in-domain data from large generic corpora for neural machine translation.

Findings

Models trained on selected in-domain data outperform those trained on generic data.

The method achieves high-quality domain adaptation at low computational cost.

Selected data improves translation accuracy in specific domains.

Abstract

Continuously-growing data volumes lead to larger generic models. Specific use-cases are usually left out, since generic models tend to perform poorly in domain-specific cases. Our work addresses this gap with a method for selecting in-domain data from generic-domain (parallel text) corpora, for the task of machine translation. The proposed method ranks sentences in parallel general-domain data according to their cosine similarity with a monolingual domain-specific data set. We then select the top K sentences with the highest similarity score to train a new machine translation system tuned to the specific in-domain data. Our experimental results show that models trained on this in-domain data outperform models trained on generic or a mixture of generic and domain data. That is, our method selects high-quality domain-specific training instances at low computational cost and data size.