Deceptive Opinions Detection Using New Proposed Arabic Semantic Features

Some users try to post false reviews promote or devalue other’s products and services. This action is known as deceptive opinions spam, where spammers gain profit from posting untruthful reviews. Therefore, we conducted this work develop implement new semantic features improve the Arabic deception detection. These were inspired study of discourse parse rhetoric relations in Arabic. Looking importance phrase unit language grammatical studies, have analyzed selected most used markers calculate proposed features. last basically represent texts classification phase. Thus, accurate technique area which has been proven by several previous works Support Vector Machine classifier (SVM). But there always a lack concerning annotated resources specially for detection it considered research area. semi supervised SVM overcome problem using unlabeled data.