Unsupervised Parallel Sentences of Machine Translation for Asian Language Pairs

Abstract

Parallel sentence pairs play a very important role in many natural language processing (NLP) tasks, especially cross-lingual tasks such as machine translation. So far, many Asian language pairs lack bilingual parallel sentences. As collecting bilingual parallel data is very time-consuming and difficult, it is very important for many low-resource Asian language pairs. While existing methods have shown encouraging results, they rely on bilingual data seriously or have some drawbacks in an unsupervised situation. To address these issues, we propose a new unsupervised similarity calculation and dynamic selection metric to obtain parallel sentence pairs in an unsupervised situation. First, our method maps bilingual word embedding (BWE) by postdoc adversarial training which rotates the source space to match the target without parallel data. Then, we introduce a new cross-domain similarity adaption to obtain parallel sentence pairs. Experimental results on real-world datasets show that our model can obtain better accuracy and recall on mining parallel sentence pairs. We also show that the extracted bilingual sentence corpora can significantly improve the performance of neural machine translation.

Chinese–Vietnamese Pseudo-Parallel Sentences Extraction Based on Image Information Fusion

“…XiaYang et al [9] effectively improved the quality of extracted sentence pairs by using a very small seed lexicon (about hundreds of entries) during the process of learning cross-lingual word representations. ShaoLin et al [10] proposed a new unsupervised method for obtaining parallel sentence pairs by mapping bilingual word embeddings through postdoc adversarial training and introducing a new cross-domain similarity adaption. YuSun et al [11] proposed an approach based on transfer learning to mine parallel sentences in an unsupervised setting, which utilizes bilingual corpora of rich-resource language pairs to mine parallel sentences without bilingual supervision of low-resource language pairs.…”

“…These methods relied heavily on parallel data, which is unsuitable for low-resource scenarios. While transfer learning and unsupervised learning based on cross-lingual word embeddings and multilingual pre-trained models are the current mainstream research direction for low-resource languages [8][9][10][11][12][13], they may not be effective for languages with substantial differences.…”

Development and Validation of Parallel Corpus: A Framework for Building Bikol-Filipino Linguistic Resource

“…They are particularly useful for machine translation, where the goal is to automatically translate text from one language to another. However, despite their usefulness, parallel corpora are still lacking in many Asian languages [4], which poses a challenge for researchers and developers working on improving multilingual processing. Oco and Roxas have highlighted the issue of insufficient resources as a major setback in research on Philippine languages [5].…”

Realization of Chinese-English Bilingual Speech Dialogue System using Machine Translation Technology

“…This data can come in various forms, such as text, images, or numerical values, depending on the task at hand. Next, the data undergoes preprocessing, where it is cleaned, organized, and transformed into a format suitable for analysis [7]. This step may involve handling missing values, normalizing features, or encoding categorical variables [8].Once the data is prepared, a machine learning model is selected and trained using the preprocessed data.…”