ML Approach For Detecting Suspicious Digital Interactions

Abstract

The rapid expansion of digital communication platforms and online services has significantly increased the risk of cyber threats, particularly phishing attacks and malicious web interactions. Attackers frequently exploit deceptive URLs and fraudulent websites to obtain sensitive information such as login credentials, financial data, and personal details. Traditional security mechanisms based on rule-based filtering and blacklists often struggle to detect newly emerging threats due to the constantly evolving nature of cyberattacks. To address these limitations, this study proposes a machine learning-based approach for detecting suspicious digital interactions through phishing URL classification.The proposed system analyzes structural and behavioral characteristics of URLs and applies machine learning techniques to classify them as legitimate or malicious. Several supervised learning algorithms, including Random Forest, Support Vector Machine (SVM), and XGBoost, were implemented and evaluated using labeled datasets. In addition, deep learning models such as Deep Neural Networks (DNN) and One-Dimensional Convolutional Neural Networks (CNN-1D) were employed to capture complex patterns and sequential relationships within the data. The dataset was preprocessed through feature extraction, normalization, and data cleaning techniques to improve model performance.Experimental results demonstrate that both machine learning and deep learning models are capable of effectively detecting phishing URLs with high accuracy. Among the evaluated models, ensemble learning techniques and deep learning architectures showed superior performance in identifying malicious patterns. Furthermore, a user-friendly web application was developed to allow users to input URLs and receive real-time predictions regarding their safety. The results indicate that the proposed approach provides a scalable and efficient solution for improving cybersecurity by automatically detecting suspicious digital interactions. Future enhancements may include integrating larger datasets, real-time threat intelligence, and advanced deep learning architectures to further improve detection accuracy and system adaptability.