Named Entity Recognition For The Classical Languages For The Building Of A Catalog Of Ancient Peoples

FAIN: HAA-256078-17

Ohio State University (Columbus, OH 43210-1349)
Brian Daniel Joseph (Project Director: January 2017 to October 2021)
Christopher Brown (Co Project Director: May 2017 to October 2021)
Micha Elsner (Co Project Director: May 2017 to October 2021)
Marie-Catherine de Marneffe (Co Project Director: May 2017 to October 2021)

The creation of a catalog of individuals and groups of individuals mentioned in ancient sources, in part to focus attention on the historical role played by those other than the “great actors” (the important individuals, states, or empires singled out in historic texts). To do so, they will use Named Entity Recognition, a computational linguistics method which identifies people and place names in texts and then sorts them into pre-defined categories, allowing further study and analysis.

The Herodotos Project is creating a catalog of all groups of peoples mentioned in ancient sources, ultimately to assemble informational material for a detailed ethnohistoric profile of each. Our sources at first are Latin and Greek texts. Given the labor-intensive and time-consuming nature of manually searching texts in the original language, and for greater accuracy, we are automating the group name extraction process, drawing on Named Entity Recognition (NER) technology from computational linguistics to identify significant entities in a given text, including our target group names. Most NER systems are English-based, so we have been creating a Latin system that is successful (c. 90% accuracy) but needs more development to achieve even better results. Also, we must adapt our Latin-based system for use with Greek. The NER-development phase of the Project is an essential step towards furthering the creation of the catalogue that will fuel the ethnohistoric side of the overall project.

Associated Products

Practical, Efficient, and Customizable Active Learning for Named Entity Recognition in the Digital Humanities (Article)
Title: Practical, Efficient, and Customizable Active Learning for Named Entity Recognition in the Digital Humanities
Author: Alexander Erdmann
Author: David Joseph Wrisley
Author: Benjamin Allen
Author: Christopher Brown
Author: Sophie Cohen- Bode´ne`s
Author: Micha Elsner
Author: Yukun Feng
Author: , Brian Joseph
Author: Be´atrice Joyeux-Prunel
Author: Marie-Catherine de Marneffe
Abstract: Practical, Efficient, and Customizable Active Learning for Named Entity Recognition in the Digital Humanities
Year: 2019
Primary URL: https://cpb-us-w2.wpmucdn.com/u.osu.edu/dist/4/27964/files/2019/08/2019_NAACL_HER_OL.pdf
Secondary URL: https://u.osu.edu/herodotos/presentations/
Format: Journal
Periodical Title: Proceedings of North American Association of Computational Linguistics (NAACL 2019)

The Herodotos Project: Towards an ethnohistory of the ancient world (Article)
Title: The Herodotos Project: Towards an ethnohistory of the ancient world
Author: Christopher Brown
Author: Marie-Catherine de Marneffe
Author: Micha Elsner
Author: Brian D. Joseph
Abstract: Towards an ethnohistory of the ancient world
Year: 2018
Primary URL: https://digitaleditions.glpublishing.com/pathwayswinter2018/
Primary URL Description: See p. 20ff. in electronic version.
Format: Magazine
Periodical Title: PATHWAYS. A Publication of Ohio Humanities
Publisher: Ohio Humanities (Winter 2018), 20-23.

---