What Led Us to the Data

Abstract

The usefulness of Bibles in Natural Language Processing is wildly underrated, especially when looking at low-resource languages. The reason these texts are so useful in machine translation is because it provides parallel text in languages that would otherwise have non or very little other parallel texts available. The question is then, how can we turn these parallel texts into a useable, organized, and reliable data set? There are two general methods that can be used to collect and organize this data, the old fashioned way by hand or automating the process through a computer. While this paper will touch on both methods, it takes a much deeper dive in the automated processing side and looks at some of the reasons this route was chosen.