Skip to main content

Detect and monitor extremist rhetoric or planned criminal activities using social media and dark web multimodal data

Primary supervisor

Thanh Thi Nguyen

Research area

Vision and Language

This project aims to employ advanced machine learning techniques to analyse text, audio, images, and videos for signs of harmful behaviour. Natural language processing algorithms are utilized to examine vast amounts of textual data, identifying keywords, phrases, and sentiment that may indicate extremist views or intentions. Analysing audio involves techniques such as speech recognition, keyword analysis, contextual analysis, audio feature extraction, and machine learning models to identify and assess potentially dangerous content. Similarly, computer vision models are implemented to analyse images and videos for symbols, gestures, or contextual clues associated with radical groups or criminal activities. Multimodal approaches that integrate text, audio, and visual data enhance the detection capabilities, enabling systems to draw more comprehensive insights from various sources. This project relies on the collaborative learning between the models to improve the detection accuracy. Additionally, the developed methods can help identify emerging trends and patterns in rhetoric or planning activities, allowing for timely intervention by authorities. These monitoring systems are essential for public safety, enabling proactive measures against potential threats.

Required knowledge

Python programming

Machine learning background


Learn more about minimum entry requirements.