Understanding Online Hotel Reviews Through Automated Text Analysis
Shawn Mankad, Spring H. Han, Joel Goh +1 more
Customer reviews submitted at Internet travel portals are an important yet underexplored new resource for obtaining feedback on customer experience for the hospitality industry. These data are often voluminous and unstructured, presenting analytical challenges for traditional tools that were designed for well-structured, quantitative data. We adapt methods from natural language processing and machine learning to illustrate how the hotel industry can leverage this new data source by performing automated evaluation of the quality of writing, sentiment estimation, and topic extraction. By analyzing 5,830 reviews from 57 hotels in Moscow, Russia, we find that (i) negative reviews tend to focus on a small number of topics, whereas positive reviews tend to touch on a greater number of topics; (ii) negative sentiment inherent in a review has a larger downward impact than corresponding positive…