This pioneering research evaluates how Large Language Models (LLMs) can automatically identify and categorize violence in ancient texts, transforming manual historical research into an efficient automated process.

• Reduces time-intensive manual harvesting of violence-related data from historical texts
• Enables historians to analyze societal dynamics surrounding conflicts more efficiently
• Provides a scalable approach to processing large volumes of ancient literature
• Creates new opportunities for computational analysis in humanities research

For security applications, this research demonstrates how AI can identify concerning content in large text corpora, potentially enabling more effective threat assessment methodologies and violent content monitoring systems.

Automating Violence Detection and Categorization from Ancient Texts