Researchers have developed two new methods for aligning words across entire documents in different languages. The methods, CTFAlign and MDPAlign, can improve the accuracy of word alignment by up to 21% compared to existing approaches. This is particularly useful for tasks such as machine translation, where accurate word alignment is crucial. The methods operate directly on full documents without relying on sentence segmentation or alignment, making them more efficient than previous techniques. --- Why it matters: This matters because it can improve the accuracy of downstream applications like document-level translation and semantic difference recognition. By reducing word alignment error rates, these methods can lead to better performance in tasks that rely on accurate cross-lingual understanding. Source: https://arxiv.org/abs/2608.21023