The on-demand economy has prospered with the rapid development of digital platforms. Many customers use on-demand service platforms to order services and then post online reviews. Using text-mining approaches, this study examines customers' online review-writing behavior and their overall satisfaction with restaurants in the context of on-demand food service. We use customers' overall ratings in their reviews to measure their overall satisfaction. We find that customers comment on the main service provider, the restaurants, and the auxiliary service providers, which include the drivers and on-demand service platform; in their online customer reviews, which are posted on the restaurants' web pages on the on-demand service platform. From text regressions, we found the determinants of customer satisfaction with the restaurants through their online reviews. There is a spillover effect from the performance of auxiliary providers on customer satisfaction with the main provider. That is, the performance of the drivers and the platform affects customers' overall satisfaction with the restaurants. In addition, we find that a higher cost of the order makes customers comment more on the attributes offered by the restaurants to show their overall satisfaction. Further, we find the type of listed merchants categorized by their properties (i.e., chain or independent) and participation in the platform programs affect the influence of the various attributes, offered by different providers, on customer satisfaction with the main provider. The findings shed light on the determinants of customers' overall satisfaction and urge improvement in collaboration and coordination between various participants in the on-demand service context.

随着数字平台的快速发展，按需经济蓬勃发展。许多客户使用按需服务平台订购服务，然后发布在线评论。本研究使用文本挖掘方法，研究了顾客在按需餐饮服务中的在线评论撰写行为以及他们对餐馆的总体满意度。我们在评论中使用客户的总体评分来衡量他们的总体满意度。我们发现客户评论了主要服务提供商，餐厅和辅助服务提供商，其中包括驱动程序和按需服务平台；在他们的在线客户评论中，这些评论发布在按需服务平台上的餐厅网页上。根据文字回归 我们通过在线评论找到了客户对餐厅满意度的决定因素。辅助提供商的绩效会对客户对主要提供商的满意度产生溢出效应。也就是说，驱动程序和平台的性能会影响客户对餐厅的整体满意度。此外，我们发现，较高的订单成本使顾客对餐厅提供的属性发表更多评论，以显示其总体满意度。此外，我们发现按属性（即连锁或独立）分类的列出商家的类型，参与平台程序会影响不同提供商所提供的各种属性对主要提供商对客户满意度的影响。调查结果揭示了客户的决定因素