Fake review detection in e-Commerce platforms using aspect-based sentiment analysis

Consumers rely on internet user reviews. Existing sentiment-based detection systems fail to capture consumer feelings regarding numerous aspects of products or services which influence their purchasing decisions. Despite the growing interest in detecting false reviews, prior studies have not explored the capacity to detect fake reviews for diverse products, which require distinct consumer experience. To overcome these problems, this paper proposes a fake review detection model using aspect-based sentiment analysis (ABSA) while considering the effects of product types. Using a dataset of Amazon reviews, our ABSA model revealed that two aspects are fundamental for detecting fake reviews and suggests the need to associate the two. These are the product category and the verified purchase attribute (with the greatest contribution observed for credence and experience product types).

Automated summary

Consumers rely on internet user reviews, but current sentiment-based detection systems are inadequate in capturing consumer feelings regarding various aspects of products or services. This study proposes a fake review detection model using aspect-based sentiment analysis (ABSA) that considers the effects of product types. The ABSA model, applied to a dataset of Amazon reviews, identified product category and verified purchase attribute as fundamental in detecting fake reviews, particularly for credence and experience product types.