An Intelligent Model for Text Forensics from Arabic Twitter Posts

Abstract

The number of twitter users in Arab countries at a rapid pace of growth. As for the social networking platforms, they are a suitable environment for luring innocent users and commit different forms of crimes against them by cyber predators through claiming something they are not such as fake ID, age or gender targeting certain segments of the society. This have resulted in a huge number of cybercrimes including Phishing, harassment, cyber terrorism and many other forms. In this context, a number of different research efforts in the literature are available for English but very rare for Arabic texts. This article proposes an intelligent technique for enhancing text preprocessing that impacting the accuracy and performance of the classification task. The proposed model was evaluated against recently collected dataset from Arabic twitter users using different classifiers including naïve bays and neural networks to predict age attribute.