A Novel Node Similarity Measure for Efficient Email Classification

Electronic mail emerged as a key communication medium even after various alternatives, such as social networks, mobile applications, and electronic messages. Continuous increase in the number of email users leads to an increase in unwanted emails for any individual, necessitating auto management. Auto-management of email is an automatic email system that classifies emails into a predefined category based on training. This paper presents a practical e-mail classification approach based on node similarity measures. The proposed method has three major phases: building a graph for each email where attributes of the email are represented as nodes of the graph with edges connecting between two nodes, finding the similarity between two graphs with the help of the Node Similarity method, and finally classifying emails based on the similarity between the graphs. The paper did a comparative analysis with related works to assess the performance of the proposed approach. The noticeable results were obtained by the presented method.