Affected by the factors like population, economic and geographic conditions, accommodation offerings are inherently different at different cities and characterized with their specific features. This heterogeneity is not limited to the supply-side but covers the demand-side as well. For instance, business and leisure tourists may favor different travel destinations. Therefore, the development of accommodation industry needs to match the evolving demands of tourists. In this study, we utilize text mining techniques to understand English-speaking tourists’ likes and dislikes with regard to hotels at different Chinese cities. Based on the studying the titles of 96,089 English reviews collected from TripAdvisor, the study seeks to explore the hotel attributes that tourists discussed in their reviews with regard to their lodging experience at a specific city. In particular, tourists’ complaints are studied through the use of low rating reviews. In addition, hotel features favored by tourists are identified via associating hotel attributes with sentiment-featured words like ‘great’ and ‘good’. Finally, an overall perspective on customer reviews is visualized in co-occurrence maps of jointly used sentiments and key-words. The research findings offer city-level strategic insights for hotel management.