An Unsupervised Approach for Content-Based Clustering of Emails Into Spam and Ham Through Multiangular Feature Formulation

The rapid growth of spam email attacks and the inherent malicious dynamism within those on a range social, personal business activities warrants an intelligent automated anti-spam framework. Attempts like malware propagation, identity theft, sensitive data pilfering, monetary as well reputational damage are sharply increasing, endangering privacy victim. Current solutions that rather incomplete when multidimensional feature email, is taken into account. We believe methodology based Artificial Intelligence, especially unsupervised machine learning way forward. This research attempts to investigating application for clustering Spam Ham emails. overall goal develop framework solely depends methodologies through approach includes multiple algorithms, primarily using content (body) subject header. has been done novel binary dataset 22,000 entries ham emails, composed ten features (reduced from eleven after reduction). Seven out these unique this study, engineered represent impactful analytical characteristics multiangular point view. Out five different algorithms investigated in work, OPTICS produced optimum demonstrating 0.26% higher average efficacy than its nearest performer DBSCAN. balanced accuracy DBSCAN was found be ≈75.76%.