A13: Aptimas Advanced Automated Approach Aimed At Author-Artifact Attribution And Assessing Author Ambitions

Human social interaction is complicated by ambiguous identities, misinformation, misdirection, and flexible alliances. Additionally, thanks to the internet, authors can post messages from anywhere in the world, in diverse forms, and with different personas. Current solutions are challenged to accurately identify and characterize authors across platforms and data types, let alone infer authorial intent. To address this shortcoming, Aptima and its partners at The Pennsylvania State University Applied Research Laboratory propose a system to aid the intelligence analyst in identifying actors and their ambitions over time, data type, and domain. The system will rely on an ensemble of automated algorithms and techniques in machine learning, information theory, and algebraic topology to extract features from digital artifacts and match those features to one another by similarity metric. Interactive graphical and tabular representations will make authors, intents, and their intersection comprehensible. The system will integrate with other systems across domains in real-time; its modular structure will allow the addition and enhancement of functionality with minimal, if any, disruption to downstream pipelined processes. In summary, the system is intended as an essential tool for those whose business depends on understanding the implications of human social interaction, despite its vagaries.