Opinion Fraud Detection in Streaming Comments Utilising Node Similarity in the Review Network

Due to the exponential growth of electronic commerce, online shopping has become a part of our lives. One important criterion in decision making when we want to purchase a product or a service is users’ reviews. When something is valuable, it’s fake will be created as well. It is the same for users’ reviews. The purpose of these fake reviews is to deceive users, leading them to make a wrong choice. One challenging problem is when we can trust a review. Although many researchers attempted to address this problem, none of them pictured the problem in a streaming domain. With the help of the review network’s properties, we propose a model to find reliable reviews when reviews are coming in a stream. Our model is fast and online, that is, it is capable of identifying reliable reviews as they are been submitted, and scalable because it is acomplementary model to offline models in detecting fake reviews.

Keywords:

big data , opinion fraud , network science , review spam , spark , fake reviews , streaming data