Detect fake identities using improved Machine Learning Algorithm

Abstract

Nowadays, Social networks are controlling the world. Number of users using the social media platforms are continuously increasing every year. The main advantage of Online social media is anyone from anywhere in the world can connect to any users easily and able to communicate with them. This paved the way of malicious attacks like fake identities and spams. The Survey of Facebook and Twitter states that the actual users present in their platforms is lesser than the number of accounts created. It means there is an increase in fake profiles over the year. With these accounts, their creators can influence the users and distribute false information. They exploit the users in order to achieve their hateful aims, for example, creating hateful posts/tweets. The increasing number of fake accounts in social media platforms is becoming a major risk for the internet users. So, there is a need for identifying these fake identities. The existing machine learning algorithms are SVM linear, naïve bayes, random forest and Decision tree algorithm which are used to detect fake identities provides less accuracy and has become inefficient. We implement Ada-boosting algorithm with SVM to detect and identify bots. The proposed model is Ada-boost SVM classifier provides higher accuracy in detecting the fake accounts and show better results in the detection of identity bots.