Profile Similarity Recognition in Online Social Network using Machine Learning Approach

Now-a-days., On-line Social Networks (OSN) plays a major role. It helps to maintain a social bond across world. People can share their information like personal, political and other career related work within minutes. High number of people started indulging in social networks because of its time complexity and easiest way of sharing details. On the other hand, the threats on social networks increased in a rapid way to gain sensitive information of a people sharing important details like passwords, bank details, personal details etc. The online privacy and security threats reports gone to top. It made researchers to involve more on this area of threats in order to protect individual privacy and security regarding social networks. Attackers started to exploit the vulnerabilities of OSNs which highly attracted malicious entities. Each OSNs had ten to hundred million users collectively generate billions of personal sensitive data content. To protect users from exploitation is one of the major challenges. The current research are mainly focused to protect the privacy of an individual's online profile in OSN. The creator constructs a fake profile to pretend himself as a genuine person in these sites. The aim in this paper is to check and detect the trusted and fake profiles created in OSN through machine learning based techniques. By using algorithms like (SVM)Support vector machine, (DNN) Deep neural network, Random Forest to overcome processes such as treating missing value, identification of variable, cleaning and validation of data will be done on the entire dataset. Here, python language is used for improving the efficiency of code. With help of this, millions of fake profiles can be detected automatically. Additionally, various machine learning algorithms are compared and discussed along with the classification report. By identifying the confusion matrix, the results show the effectiveness of the proposed machine learning algorithm, which has best accuracy, precision, Recall and F1 score.