On online shopping platforms, product reviews matter given their impact on consumers' shopping decisions and product sales. However, a large number of fake reviews are created to lure customers to buy products, which brings poor shopping experiences to customers and further damages the platform's reputation. Prior studies mainly leveraged textual features to detect fake reviews, failing to deal with multi-modal reviews with images. Therefore, to bridge this gap, we propose a new end-to-end framework to detect fake reviews in e-commerce platforms. We first design flexible encoders for unimodal feature extraction. Then we imitate the human browsing behaviors by designing a co-attention mechanism. The co-attention explores the dependency between textual and visual modalities, thus further enhancing the performance of fake review detection. Experiments conducted on a real-world dataset demonstrate the effectiveness of our model.


Paper Number 1782; Track Design; Complete Paper



When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.