Skip to content
/ DLT2 Public

Dating Latin Texts with Deep Learning Techniques

Notifications You must be signed in to change notification settings

Sittch/DLT2

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

In this paper, we introduce DLT2, a model that can date Latin texts through deep learning techniques in order to assist archaeologists, classicists, and other specialists. Latin is a classical language with a long, rich history, and as such, there are many significant texts for which a date cannot be determined precisely, if at all. The model is trained on a corpus of Latin texts with dates of writing that can be estimated within a 50-year window. These texts are divided into eight distinct "eras" based on recognized developments in the historical development of Latin and its societal contexts. The model is then able to suggest the most likely era for an undated Latin text. After ensuring that metadata such as dates, authors’ names, and titles are removed from the beginning of each document, DLT2 is able to achieve an overall prediction accuracy of 30.88% in classifying texts with known dates, with some era-specific accuracy scores reaching as high as 66.85%.