Multinomial Naive Bayes Classification AlgorithmBased Robust Spam Detection System
Shivansh Rastogi, Rakesh Singh Sambyal, Priyanka Tyagi, Ritik Kushwaha · 2024
In the age of ubiquitous digital communication, the steady stream of unsolicited emails presents a serious obstacle to effective and secure communication. In this research work, a comprehensive Spam Detection System based on a novel method based on the Multinomial Naive Bayes (MNB) Classification Algorithm is presented. The system starts by carefully cleaning and standardizing raw email data through text preprocessing. Next, using Count Vectorizer to convert the processed text into numerical features, the frequency of words in each email is recorded. The Multinomial Naive Bayes classifier that follows is built on top of this token count matrix. The Multinomial Naive Bayes model, which is well-known for its effectiveness when working with big feature sets, uses the training data to identify trends and categorize emails as either legitimate or spam. The exceptional accuracy rate of 99.67% achieved by our spam detection system is proof of the effectiveness of the Count Vectorizer and Multinomial Naive Bayes pipeline working together. The project places a strong emphasis on continuous learning, which enables the system to update frequently in response to changing spam tactics.