Machine Learning in Textual Criticism: An examination of the performance of supervised machine learning algorithms in reconstructing the text of the Greek New Testament

New Testament textual criticism is a field of research that seeks to confidently establish the text of the New Testament by comparing thousands of manuscripts, finding the variants between these manuscripts, and making informed decisions as to the original text based on the internal and external features of those manuscripts. By examining such features as date of composition, textual family, and variant length, scholars are able to determine the correct reading of a text with a high degree of confidence. The use of computing in this field has been documented since at least the 1970’s, but they have not been applied to the task of textual criticism itself. Rather, computers have been used primarily to classify manuscripts and determine their relationships to each other. Our research in this paper takes a new approach by applying machine learning algorithms to the task of textual criticism. After testing multiple supervised learning algorithms, our research finds that support vector machines and decision trees outperform the other tested and achieve 98.8% accuracy.