Detection of Fake Profile on Social Media Using Machine Learning and Feature Selection Techniques
Social media facilitates the sharing of not only feelings and expressions but ideas and information too. Meanwhile, when someone attempts to clone one's profile with a malevolent intention, it not only breaches its privacy but can sabotage in other senses too. Many researchers in the past have endeavoured to prevent these kinds of malicious pursuits on the internet. With the help of Machine Learning (ML) and Feature Selection techniques, fake profiles can be detected at an introductory stage so that one is not capable of performing scurrilous efforts on the site. This work is an endeavour to detect whether a profile is fake or not based on users profile information. In this research, a model is proposed based on data pre-processing, feature selection and ML techniques to detect fake profile. The data pre-processing encompasses the measures like removing null values, encoding, and feature scaling. Feature selection is performed as an endeavour to reduce the dimension of the dataset and avoid overfitting. After applying feature selection, the number of features gets reduced from 34 to 11. Seven single ML techniques are employed to evaluate the effectiveness of data-pre-processing and reduced 11 features to detect fake profile. The results evince that data pre-processing and feature selection techniques improve the accuracy, precision, recall, and F1-Score of ML techniques and hence the performance of model.