Adaptative Bilingual Aligning Using Multilingual Sentence Embedding

TL;DR

This paper introduces AIlign, an adaptive bilingual alignment system using multilingual sentence embeddings that effectively aligns fragmentary and non-monotonic texts, matching state-of-the-art performance with improved flexibility.

Contribution

AIlign is a novel alignment method leveraging sentence embeddings for adaptive, non-monotonic, and fragmentary bilingual text alignment, outperforming recent systems in flexibility.

Findings

Achieves state-of-the-art alignment accuracy

Handles non-monotonic and fragmentary texts effectively

Maintains quasi-linear computational complexity

Abstract

In this paper, we present an adaptive bitextual alignment system called AIlign. This aligner relies on sentence embeddings to extract reliable anchor points that can guide the alignment path, even for texts whose parallelism is fragmentary and not strictly monotonic. In an experiment on several datasets, we show that AIlign achieves results equivalent to the state of the art, with quasi-linear complexity. In addition, AIlign is able to handle texts whose parallelism and monotonicity properties are only satisfied locally, unlike recent systems such as Vecalign or Bertalign.