A Hybrid Approach to Trace and Mitigate Attacks On Cyber cased Objects in Social Networks Using SVM and Smote

Abstract

With the growing reliance on location-based services and the widespread newlineadoption of GPS-enabled devices, developing systems that can effectively retrieve newlineand study geotagged objects using GPS parameters is becoming increasingly newlineimportant. newlineExisting security measures are insufficient to preserve the privacy and are newlineineffective for obtaining proof against cyber-attacks that are continually evolving in newlinethe web arena. To safeguard his/her reputation and resources kept in the social newlinenetworks, the user s privacy should be given top priority. To make the forensic newlinepowerful enough to detect and present solid evidence in a court of law, an advanced newlinecybercrime architecture and algorithms are required. newlineThe following are some of the hurdles that must be overcome by the user in newlineorder to avoid complications with geotagged objects. newline(1) At the organizational level, there is a lack of awareness about cyber cased objects. newline(2) There were no communication policies in place when it came to geotagging of newlineobjects. newline(3) In the case of dynamic objects, there was a failure to forecast the changes. newline(4) There are no established standards for dealing with cyber cased objects. newlineThe proposed system is implemented in Visual C++ and python. The UI is run newlineon a machine with an Intel Core i5-7200 processor operating at 2.7 GHz and 8 GB of newlineRAM. The KDD Cup dataset was used for this research, which is a widely used newlinedataset in the field of intrusion detection. To evaluate the proposed model, we first newlinetrained it on the dataset and then performed testing. During the evaluation process, we newlineconsidered several performance metrics, including accuracy, precision, sensitivity, newlinespecificity, and F-Score, for both training and testing phases. We split the dataset into newline70% for training and 30% for testing, ensuring that the model was trained and tested newlineon different sets of data. This approach helped us to avoid over fitting and ensured newline