The fit information of multi-modal user-generated content has its unique value but is understudied. This study investigated how the fit between photos and text within a hotel review influences tourist rating behavior. Employing deep learning technologies and econometric analysis shows that the fit between photos and text has an inverted U-shape effect on tourists’ ratings using online hotel review data from Qunar.com. The dispersion of ratings and the number of reviews can weaken the above effect; this is due to the varying level of cognitive resources tourists possess. Endogeneity bias from text and photos is excluded by conducting the two-stage residual analysis. The information uncertainty reduction mechanism is tested between photo-text fit and ratings. This paper provides valuable insights for practitioners regarding user-generated content management in hospitality industry.

中文翻译：

---

文字与照片相遇：视觉内容如何影响评级


多模态用户生成内容的拟合信息有其独特的价值，但尚未得到充分研究。这项研究调查了酒店评论中照片和文本之间的契合度如何影响游客评分行为。采用深度学习技术和计量经济学分析表明，使用 Qunar.com 的在线酒店评论数据，照片和文字之间的拟合对游客的评分具有倒 U 形影响。评分的分散和评论的数量会削弱上述效果;这是由于游客拥有的认知资源水平不同。通过进行两阶段残差分析，排除了文本和照片的内生性偏倚。在照片文本拟合和评级之间测试了信息不确定性降低机制。本文为酒店业用户生成内容管理的从业者提供了宝贵的见解。