Data Science Analysis of Malicious Advertisements and Threat Detection Automation for Cybersecurity Progress
Sandra Nguyen, Doina Bein · 2023
We live in an era of unprecedented technology. Millions of users depend on information technology to carry out their daily lives and large-scale commercial and industrial operations are no exception. At the same time, the rapidly growing interconnectivity of IT systems and the surge in cybercrime since the pandemic have rendered industry-standard hardware and software components increasingly vulnerable to malicious attacks. Cyber defense is a coordinated act of resistance that intends to understand the capabilities and motives of attackers in order to secure our country's data and more importantly, the livelihoods of our citizens. This research aims to contribute to the progress of cybersecurity and defense technology as a whole by focusing on a dynamic aspect of malware: digital unwanted advertisements. It presents a novel approach to automating the analysis of malicious content on the internet by web scraping ads of the popular search engine Google to extract relevant data (URL, Company, Title, Product Desc.), building machine learning models (supervised & unsupervised) to classify and make predictions on that data, and creating a web application for end users to access. The results show that our tool can detect trends within the features with limited false positives, paving the way for us to make predictions on whether the advertisements are desirable or unwanted. The research concludes that in this time and age, it is extremely important to protect against fraud, especially by adhering to cybersecurity's best practices and to think about threats in more global terms. Our hope with this research is to prompt action to ensure society continues to improve in IT resilience.