Disinformation on social media poses significant detrimental effects, and also presents unique challenges due to the large scale, diverse topics, annotation cost, and debunking obstacles.
IDDP research fellow, Dr. Kai Shu, aimed to understand disinformation with social context and propose a systematic computational framework with three coherent tasks:
(1) Detection: bringing multi-modal social context as auxiliary information for disinformation detection with limited supervision;
(2) Adaptation: transferring knowledge across domains with adversarial machine learning; and
(3) Explanation: discovering useful explanation from social media without sacrificing detection performance.
MEETING PASSCODE IS: 557181