9th International Conference on ICT for Sustainable Development

Authors - Gayatri Joshi, Soham Joshi, Vatsal Mehta, Kumkum Saxena
Abstract - Emails have become a centre for a large user base, leading it to be a platform for reaching large crowds with minimum effort and enabling an effortless spread of information to larger audiences. But on the downside, it has also led to an increase of user traffic in email, which also consists of some unwanted marketing and promotion-related emails. Several users' private data is being extracted, and such superfluous emails are sent which are largely known as spam. The paper aims to combat the problems by making use of different machine learning approaches. Despite the multitude of available classifiers to build classification models for an imbalanced dataset, the main task was to reduce the bias of the majority class. This paper aims to restructure the imbalanced nature of the dataset and apply different machine learning algorithms, like Support Vector Machine, Random Forest, Multinomial Naive Bayes and XGBoost. In addition to this, the paper also compares between different ensemble techniques used. Not only does this avoid erroneous categorisation of legitimate emails as spam, but also provides insightful knowledge to rectify imbalanced dataset problems. It underscores the potential of using different methods to enhance the accuracy and combating the problem.