Publication

Abstract : In the era of social networking and e-commerce sites, users provide their feedback and comments in the form of reviews for any product, topic, or organization. Due to high influence of reviews on users, spammers use fake reviews to promote their product/organization and to demote the competitors. It is estimated that approximately 14% of reviews on any platform are fake reviews. Several researchers have proposed various approaches to detect fake reviews. The limitation of existing approaches is that complete review text is analysed which increases computation time and degrades accuracy. In our proposed approach, aspects are extracted from reviews and only these aspects and respective sentiments are employed for fake reviews detection. Extracted aspects are fed into CNN for aspect replication learning. The replicated aspects are fed into LSTM for fake reviews detection. As per our knowledge, aspects extraction and replication are not applied for fake reviews detection which is our significant contribution due to optimization it offers. Ott and Yelp Filter datasets are used to compare performance with recent approaches. Experiment analysis proves that our proposed approach outperforms recent approaches. Our approach is also compared with traditional machine learning techniques to prove that deep neural networks perform complex computation better than traditional techniques.