Investigations on email classification using machine learning and deep learning algorithms

Abstract

The vast growth of objects connected to the internet had a huge positive impact on people due to several requirements. Communication of interconnected objects raises many concerns regarding security mechanisms. Identifying a trustable source node introduces new challenges in the domain of sharing information. The research work aims to improve the safety aspect of communication in the Social IoT environment and also aims to deliver qualitative services to users through a Multi-Layered Trust Computational (MLTC) model for enhancing the security of communication. The MLTC model aims to identify trustable nodes in the network and also analyzes the quality of the provided services. In spite of solving the problems of unreliable nodes, spamming data through email raises serious concerns. Spammers utilize the e-mail medium and Online Social Network (OSN) sites to spread spam information. Spam e-mails are sent out in bulk quantities every day and these spam e-mails often have very similar characteristics seldom allowing them to be detected using the conventional techniques. The ability to identify spam should be strong enough to recognize unwanted messages and discourage spammers. Collaborative Method (CM) and text-based detection are commonly used techniques in Spam Message Detection (SMD). The proposed research utilizes a Semantic Graph Neural Network in integration with Convolutional Neural Network (SGNN-CNN). The model converts the difficulty of categorizing emails, into a problem of categorizing graphs, by projecting emails onto a graph for effective categorization. newline