Stockholm university

Research seminar: Breaking Historical Ciphers with AI in the Loop?

Seminar

Date: Wednesday 18 October 2023

Time: 15.00 – 16.30

Location: Room C307, Department of Linguistics

Beata Megyesi, Professor in Computational Linguistics at Stockholm University, will introduce you to the DECRYPT project, aiming at the systematic study of historical ciphers, secret writings with own symbol systems designed to hide the content of the message.

old handwritten cipher
The Copiale Cipher. Photo: Beáta Megyesi

The DECRYPT project is dedicated to the development of resources and tools to decipher historical ciphers using AI and a cross-disciplinary approach  involving computational linguistics, computer vision, cryptology, history, linguistics, and philology.

During the seminar Beata Megyesi will first show some typical historical ciphers from Europe from the 14th to the 18th centuries, highlighting their uniq characteristics. Following that she will outline the projects general methodology for analyzing historical ciphers with a novel set of AI tools designed to facilitate and streamline the processing, from transcription to decipherment.

The project encompasses an extensive collection of over 7,000 encrypted sources released in the DECODE database. In addition, historical texts with diplomatic transcriptions, complemented with language models for 17 European languages are released within the HistCorp collection.

To convert cipher images into a machine-readable text format, Beata and her team have developed the TranscripTool, which includes models for several alphabets and symbol systems. The TranscripTool also allows for fine-tuning  document-specific characteristics based on user feedback. Subsequently, the transcribed text serves as input for CrypTool, an application that assists in deciphering a wide range of historical ciphertexts.

The seminar will be followed by a post-seminar in the kitchenette at the Department of Linguistics, welcome!

About the speaker

Beata Megyesi is the Department of Linguistics new Professor in Computational Linguistics. Her main research areas include natural language processing and digital philology and her scholarly pursuits center around cross-disciplinary research aimed at facilitating quantitative studies utilizing AI for the humanities and social sciences. Currently, she is working on historical cryptology to analyze and break ciphers and codes.

Read more about Beata Megyesi's research here

 

If you need Swedish Sign Language interpretation for the seminar, please contact Carla Wikse Barrow or Caroline Arvidsson as soon as possible.