PJAIT Systems for the IWSLT 2015 Evaluation Campaign Enhanced by Comparable Corpora

TL;DR

This paper presents enhancements to Statistical Machine Translation systems for diverse language pairs using comparable corpora, domain adaptation, and advanced alignment and modeling techniques, leading to improved translation quality.

Contribution

Introduces novel data adaptation and alignment techniques, including comparable corpora and domain adaptation, to improve SMT performance across multiple language pairs.

Findings

Positive impact on SMT quality demonstrated by BLEU, NIST, and TER metrics

Effective use of Wikipedia-based comparable corpora for training and testing

Improved translation results through domain adaptation and advanced alignment methods

Abstract

In this paper, we attempt to improve Statistical Machine Translation (SMT) systems on a very diverse set of language pairs (in both directions): Czech - English, Vietnamese - English, French - English and German - English. To accomplish this, we performed translation model training, created adaptations of training settings for each language pair, and obtained comparable corpora for our SMT systems. Innovative tools and data adaptation techniques were employed. The TED parallel text corpora for the IWSLT 2015 evaluation campaign were used to train language models, and to develop, tune, and test the system. In addition, we prepared Wikipedia-based comparable corpora for use with our SMT system. This data was specified as permissible for the IWSLT 2015 evaluation. We explored the use of domain adaptation techniques, symmetrized word alignment models, the unsupervised transliteration models and the KenLM language modeling tool. To evaluate the effects of different preparations on translation results, we conducted experiments and used the BLEU, NIST and TER metrics. Our results indicate that our approach produced a positive impact on SMT quality.