Polish - English Speech Statistical Machine Translation Systems for the IWSLT 2013

TL;DR

This paper investigates the impact of different training configurations on Polish-English speech translation systems using TED data, focusing on data preparation, morphological features, and evaluation metrics.

Contribution

It introduces a detailed analysis of Polish data for SMT, including morphological processing and data cleaning, to improve translation quality.

Findings

Morphological processing improves translation accuracy.

Data cleaning significantly enhances system performance.

Evaluation metrics show consistent improvements across experiments.

Abstract

This research explores the effects of various training settings from Polish to English Statistical Machine Translation system for spoken language. Various elements of the TED parallel text corpora for the IWSLT 2013 evaluation campaign were used as the basis for training of language models, and for development, tuning and testing of the translation system. The BLEU, NIST, METEOR and TER metrics were used to evaluate the effects of data preparations on translation results. Our experiments included systems, which use stems and morphological information on Polish words. We also conducted a deep analysis of provided Polish data as preparatory work for the automatic data correction and cleaning phase.