This repository contains the code for the following paper: Graph Algorithms for Multiparallel Word Alignment, To appear on EMNLP 2021.
We provide two methods to improve bilingual word alignments of a target language pair, exploiting signals from other languages in a multi-parallel corpus. You can find the code for each method in its corresponding directory. The obtain the calculated word alignments, the Gold-Standard word-alignment datasets, and other data used in the paper, check out the data directory.