Online success of the brands, products, or services is dependent on the online reviews written by the costumers who share their experiences. These reviews become an essential factor in customers’ purchasing decision. Driven by the immense financial profits, some corrupt individuals or organizations deliberately post fake reviews to promote their products or to demote their competitors’ products, trying to mislead or influence customers. Therefore, it is crucial to spot fake reviews in order to maintain the credibility of online reviews. Dealing with the uncertain aspect is essential, since these reviews are issued from the human opinions which are generally uncertain and full of ambiguity due to a large number of the included fraudulent ones. Thus, this thesis proposes to handle the fake reviews detection under uncertainty. Such uncertainty is represented and managed through the belief function framework and through some evidential machine learning techniques. We propose new evidential approaches for the detection of fake reviews based on three aspects: reviews, spammers and spammer groups. Moreover, a hybrid method combining both the spammers and the group spammers categories is also performed leading to a better detection quality. To substantiate the efficiency and the impact of our proposals compared to the literature, experiments are conducted on two labelled real-world review data-sets extracted from Yelp.com.