An Investigation of the Sampling-Based Alignment Method and Its Contributions

TL;DR

This paper enhances a sampling-based alignment method for phrase translation tables by enforcing n-gram alignments and adjusting their distribution, leading to improved translation quality in statistical machine translation.

Contribution

It introduces a novel approach to increase n-gram alignments using distribution adjustments and compares merged translation tables for better translation performance.

Findings

Increased number of n-gram alignments improves translation quality.

Distribution adjustment leads to better evaluation results.

Merging tables from different methods enhances translation accuracy.

Abstract

By investigating the distribution of phrase pairs in phrase translation tables, the work in this paper describes an approach to increase the number of n-gram alignments in phrase translation tables output by a sampling-based alignment method. This approach consists in enforcing the alignment of n-grams in distinct translation subtables so as to increase the number of n-grams. Standard normal distribution is used to allot alignment time among translation subtables, which results in adjustment of the distribution of n- grams. This leads to better evaluation results on statistical machine translation tasks than the original sampling-based alignment approach. Furthermore, the translation quality obtained by merging phrase translation tables computed from the sampling-based alignment method and from MGIZA++ is examined.