Mule: Multiclass Email Classification for Forensic Analysis Using Deep Learning

Resumen

In the early 1990s of the previous centuries, as the Internet gained popularity, electronic mail grew to be a crucial means of communication. A typical user now keeps 50% of their vital data in e-mail storage, which has increased over time. Proactive data analysis is required to thwart cyber-attacks and crimes in order to interfere with cyber operations and services. Investigators now face the enormous challenge of extracting the necessary semantic information from the volume of e-mails, which is delaying the investigation process due to the continuing expansion of data communicated via e-mails. Analysis of the email's header and body is necessary to categorize the email in order to conduct email-related crime investigations. The current keyword-based approaches and filtration produce only brief emails that omit important information. We suggested a Long-Short Term Memory (LSTM) for multiclass email classification to get around the aforementioned restriction. This method is applicable to both short and large sequences of more than 1000 characters. To achieve the optimum performance, this technique concentrates on fine-tuning LSTM parameters. In this project, we created a brand-new, effective method for email classification called "EmailSinkAI." The LSTM effectively extracts useful data from email that can be utilized as evidence in forensic investigations.