Fake Profile Identification Using Machine Learning

Online social networks have permeated our social lives in the current generation. These sites have allowed us to see our social lives differently than they did in the past. Nowadays we can connect with new friends and maintain relationships with them via social and personal activities become quite easy. Online Social Networks (OSN) are contributed in all areas such as Research in all domains, Job-related areas, Technology oriented areas, Health care, and business-oriented areas, Information gathering and data collection, and so on. One of the biggest problems on these social media platforms is fake profiles. Impersonating to be someone else and causing harm and defamation to the real person or advertising or popularizing removed propaganda on someone’s name to get more benefit is the motto of such profile creators. There have been many studies regarding these fake accounts and how can they be mitigated. Many approaches such as graph-level activities or feature analysis have been taken into consideration to identify fake profiles. These methods are outdated when compared to arising issues of these days. In this paper, we proposed a technique using machine learning for fake profile detection which is efficient. . The benchmark data set is collected and mixed with manual data first furthermore; a data cleaning technique is used to present the data more feasibly. Then the preprocessed data is used for model building with sufficient information such as profile name, profile ID name, number of followers, and so on. We added Cross validation process where many training algorithms are implemented on the given data and are then tested on the same data. Based on the experiments the RF classifier performed better than the other classification methods. The Random Forest classifier is used to forecast the profile whether is fake or genuine in an efficient way.