Is a picture worth a thousand words? Understanding the role of review photo sentiment and text-photo sentiment disparity using deep learning algorithms

Images have become integral to consumers' sharing of consumption experiences due to their abilities of carrying rich and vivid information. This study investigates the impacts of restaurant review photo sentiment on customers’ perceived review usefulness and enjoyment using deep learning and econometric model analysis. The results indicate that (1) reviews with photos are more useful and enjoyable than reviews without photos; (2) a U-shaped relationship exists between review photo sentiment and review usefulness, with the effect of review photo sentiment on review enjoyment being positive and linear. Moreover, the effects can be strengthened by the number of review photos while weakened by the text-photo sentiment disparity. The above findings are reinforced by a sample of restaurant online reviews written by tourists in Las Vegas. This study contributes to the electronic word-of-mouth literature as well as to the application of machine learning technologies in computer vision to tourism and hospitality research.