COMPUTATIONAL METHODS FOR DETERMINING THE SIMILARITY BETWEEN ANCIENT GREEK MANUSCRIPTS. Dunn, Eddie, 2012. Capstone Paper, University of North Carolina Wilmington.
The computational piece of this project utilizes data extracted from the work of B. Daniels and George Zervos. Their unpublished dissertations from Duke University present two separate critical apparatus of 89 and 45, respectively, (all different) manuscripts of the New Testament apocryphon/pseudepigrapha book now most commonly called Protoevangelium (Proto-gospel) of James. This paper will first present this early Christian work as very unique and important document whose dating and authorship has critical implications in biblical studies. This paper will then explore computational methods of preparing the textual data that might best elicit quantifiable, statistical variances/similarities in the documents. After the data preparation, several classification and exploratory data analysis techniques will be applied that provide verification for the current expert opinions on finding groups of documents that are more similar, and hence might have come from the same “family” of documents. Because there are other versions of the Protoevangelium of James that have not received much scrutiny, it is also hoped this line of inquiry might help better classify the dating and authorship of these documents. These understandings will give us further insight in the fascinating story of this document and by corollary our understanding of early Christian literature.