104484Systems for Internet Safety: Counteracting Online Predators (SIS:COP)ClosedFeasibility StudiesInnovate UK"In 2016, NSPCC reported a 50% increase in cases of online grooming. The proposed 12-month project - SIS:COP intends to demonstrate the potential for a system embodying advanced approaches for detecting and preventing online grooming as could help to reduce this. The project will develop several key technical components for detection and prevention, and integrate these into a sophisticated prototype for real-time grooming detection. The prototype will be based around three main innovations: (i) developing our peer-reviewed and award-winning predator detection approach, independently evaluated in international competition and grounded in the Cycle of Entrapment (CoE) in Luring Communication Theory, from an offline approach into a real-time component; (ii) practical exploration of the ability of deep learning (AI) approaches to automate maintenance of detection capabilities, leading to enhancement of the existing approach; and (iii) an integration with an existing enterprise system offering for exploration and demonstration of inter-operation with state-of-the-art monitoring capabilities. We anticipate resulting systems that can demonstrate significant improvement over the present state-of-the-art in which reliance on simple lists of keywords prevails but results in high rates of false alarms and the significant risk of missed detections. By contrast, our approach evaluates the contribution of phraseological patterns over time to the emergence of predatory activities, with classification by CoE stages. CoE addresses the stages of entrapment - such as 'Isolation' and 'Deceptive Trust' - involved when predators lure victims into on-going abuse; without such classification, some interactions can seem entirely innocent. Use of CoE reduces rates of both false alarms and missed detections. A measure of success for SIS:COP would be outperforming our existing detection approach on the basis of newly discovered patterns. The project will act as a precursor to developing, and subsequently selling through licensing and partnering agreements, pluggable systems and services that detect and prevent online grooming, and provide a concrete basis for further developments for a safer internet. Key target markets that this project will support us to reach are Safeguarding/Monitoring systems (e.g. schools, chatrooms and safeguarding apps) and Enterprises - private network providers."