Following violent incidents like the "Sillim Station stabbing" in 2023, online posts predicting crimes have surged, often exploiting platform anonymity. This study proposed an AI-based system for detecting and analyzing such posts, aiming to enhance monitoring accuracy, optimize crime pattern detection, and support law enforcement collaboration.
The AI-based system was designed with five core functions: data management, analysis, user reporting, crime terminology management, and new threat alerts. It can efficiently collect and process online data using tools such as Selenium, SerpAPI, OCR, and advanced AI models such as BERT and GPT. It also integrates a citizen reporting channel to address unexpected threats. Over 91,000 posts were collected between July 2023 and May 2024, demonstrating the system's capability for rapid data collection and continuous keyword refinement.
While the system demonstrated strong potential for crime prevention, privacy concerns and legal challenges must be addressed. This pilot project lays the groundwork for broader adoption of AI-driven crime monitoring technologies.