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%.
-
Notifications
You must be signed in to change notification settings - Fork 0
Sittch/DLT2
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Folders and files
Name | Name | Last commit message | Last commit date | |
---|---|---|---|---|
Repository files navigation
About
Dating Latin Texts with Deep Learning Techniques
Topics
Resources
Stars
Watchers
Forks
Releases
No releases published
Packages 0
No packages published