科研成果详情

Introduction

Owing to the intangibility of experiential tourism products (e.g., restaurant dining), online reviews have come to play a critical role in customers' decisions (Assaker & O'Connor, 2021). Technological advances have enabled customers to upload photos when posting online reviews. Compared with plain texts, photos feature more vivid information. Images can deepen reviews' emotional intensity and enhance the persuasiveness of feedback; however, limited research in tourism and hospitality has explored the roles of user-generated photos in an online review context. To bridge this gap, our study aims to reveal the effects of review photo content diversity and image–text relevance on perceived review helpfulness as well as the boundary effects of restaurant price level.

Methods

We tested the impact of review photo characteristics on review helpfulness via econometric modeling using online review data from Yelp.com. Three hundred restaurants in Las Vegas, NV were chosen as our research sample using the stratified sampling method. All posted review text, review photos, and reviewers' information were collected. In particular, review photos (i.e., of food, drink, restaurant interior, restaurant exterior, and menu) were classified and labeled via transfer learning, based on which review photo content diversity, referring to the total number of topics/categories in a review's photos, was calculated. We then computed the relevance between review photo content and the correspondent review text using the method proposed by Shin et al. (2020).

Results/Discussion/Implications

Findings indicated that both review photo content diversity and review text–photo content relevance positively influenced customers' perceived review helpfulness. Additionally, these two effects were particularly evident for low-price restaurants. This study contributes to the literature on review helpfulness by considering both review text, review photo review, and their interaction effects. Specifically, two new review photo–related concepts, i.e., review photo content diversity and review text-photo relevance, were devised and found to influence review helpfulness. Moreover, the moderating role of restaurant price was uncovered. This study expands the application of photo analysis through machine learning algorithms in tourism and hospitality.