Improve Sentence Alignment by Divide-and-conquer

TL;DR

This paper presents a divide-and-conquer algorithm that significantly accelerates sentence alignment by leveraging bilingual embeddings and Monte Carlo simulation, achieving better accuracy and efficiency over existing methods.

Contribution

The paper introduces a novel divide-and-conquer approach that reduces sentence alignment complexity from quadratic to near O(NlogN) using external embeddings and Monte Carlo methods.

Findings

Achieves 3 F1 points improvement over Bleualign on OCR data.

Turns quadratic algorithms into near O(NlogN) with Monte Carlo simulation.

Faster than Vecalign under resource constraints.

Abstract

In this paper, we introduce a divide-and-conquer algorithm to improve sentence alignment speed. We utilize external bilingual sentence embeddings to find accurate hard delimiters for the parallel texts to be aligned. We use Monte Carlo simulation to show experimentally that using this divide-and-conquer algorithm, we can turn any quadratic time complexity sentence alignment algorithm into an algorithm with average time complexity of O(NlogN). On a standard OCR-generated dataset, our method improves the Bleualign baseline by 3 F1 points. Besides, when computational resources are restricted, our algorithm is faster than Vecalign in practice.