Neural Network-based Word Alignment through Score Aggregation

TL;DR

This paper introduces a neural network model for word alignment that uses score aggregation and a soft-margin objective, achieving improved accuracy over Fast Align on multiple language pairs.

Contribution

The paper proposes a novel neural network architecture with score aggregation for unsupervised word alignment, outperforming existing models like Fast Align.

Findings

7 AER improvement on English-Czech

6 AER improvement on Romanian-English

1.7 AER improvement on English-French

Abstract

We present a simple neural network for word alignment that builds source and target word window representations to compute alignment scores for sentence pairs. To enable unsupervised training, we use an aggregation operation that summarizes the alignment scores for a given target word. A soft-margin objective increases scores for true target words while decreasing scores for target words that are not present. Compared to the popular Fast Align model, our approach improves alignment accuracy by 7 AER on English-Czech, by 6 AER on Romanian-English and by 1.7 AER on English-French alignment.