PACIS 2021 Proceedings

Paper Type


Paper Number



Extensive research have shown the significant influence of textual contents on review helpfulness prediction, however, the review images draw little attention. Actually, the information conveyed in the review images can be either additional information (complementation effect) or similar information (substitution effect) in contrast with the textual review information. We propose a novel multimodal deep learning method to better leverage the online review texts and images and capture such interaction effect between them for their helpfulness prediction. The method firstly extracts the multimodal features using the pre-trained deep learning models and then feeds into the LSTM and attention units to learn the sequential dependency relation and importance weights. We formulize the complementation-substitution based text-image interaction effects into the loss function. Empirical experiment results on a large-scale online review dataset show the superiority of our method in terms of both prediction performance and representation learning performance.



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