Emails Classification and Anomaly Detection using Natural Language Processing
Tanvi Mehta, Renu Kachhoria, Swati Jaiswal, Sunil D. Kale, Rajeswari Kannan, Rupali Atul Mahajan · 2024
Electronic mail has established itself amongst the most popular methods of conversation for both individuals and organizations, and it has become a significant area of research to classify emails and provide users with tools for data fragmentation and connection analysis for investigative purposes. An attempt is made to identify anomalous behavior prior to the public controversy by mapping the distribution of emails sent and received by the hour. Communities of users make up most social networks. To uncover people who have the same interests and forecast their behavior, it is crucial to find these communities. Based on the terms’ significance to the entire corpus, this research suggests ways to analyze the data from 500,000 emails retrieved from the Enron email data that the Federal Energy Regulatory Commission acquired during its study into Enron’s collapse. Examination of social networks and anomaly detection are used to evaluate the emails. Additionally, Wordnets are used for in-depth research to create the word cloud. Moreover, the emails are distinguished by utilizing Support Vector Machine (SVM), Naive Bayes (NB), and Recurrent Neural Network (RNN) approaches, and their performance is evaluated using several evaluation metrics. Finally, a comprehensive analysis is conducted using the experimental findings.