Abstract
Conference Title: 2018 21st Saudi Computer Society National Computer Conference (NCC) Conference Start Date: 2018, April 25 Conference End Date: 2018, April 26 Conference Location: Riyadh, Saudi Arabia Social networks offer platforms that everyone can use freely. It gives the opportunity to share information in different ways very easily with a high level of interaction. Recently, the use of social networks has paved the way for the evolution of hidden groups which may target different aspects of consideration such as political, dogmatic, ideological, or to amplify the public speech for Twitter people through posting tweets in trending hashtags. This study aims to use data mining and text mining techniques to build an authoring classification model that can find out a link between a person and a group through his vocabulary. Our assumption is that people with similar vocabulary most probably belong to the same group. In order to test the methodology of this study, the Arabic Spammers Group is chosen as a case study of hidden groups located in Twitter especially in trending hashtags. A comparison of two classification models; Naive Bayes (NB) and Support Vector Machine (SVM) -with and without stemming- is applied. The overall performance results showed that NB model achieved higher performance than SVM model in both with and without stemming experiments.