Ranking hotels using aspect ratings based sentiment classification and interval-valued neutrosophic TOPSIS

Many a times, customers regret their decision when they book a hotel room purely on the basis of price or the hotel images available online. The customers look for additional information to substantiate their decision and this has led to the popularity of the usage of online feedbacks provided by guests towards various aspects of the hotel services. This feedback more appropriately called the electronic word-of-mouth is provided either in terms of some rating or textual comments. The numerical ratings of various service aspects of the hotels posted by guests provide a comprehensive evaluation of their sentiments and assessments on a standardized scale. Studying these sentiments is necessary in order to understand the customer needs and identify the improvement areas for hoteliers. Customers consider various alternatives and gather relevant aspect information before booking a hotel room. This involves evaluating the hotel alternatives on the basis of more than one hotel characteristics. This demands application of multi criteria decision making approach for ranking of hotels. The paper proposes a hotel ranking model based on the aspect ratings accessed from Tripadvisor website. The aspects play the role of criteria consisting of service, cleanliness, value, sleep quality, room, and location. These ratings are classified into positive, neutral, and negative sentiments, which are transformed to Neutrosophic numbers and results in the formation of interval-valued Neutrosophic decision matrix. Also, since the aspect weights are completely unknown, a non-linear programming model called maximizing deviation method is employed. Lastly, the aspect weights and decision matrix are combined to perform the procedure required for applying technique for order preferences by similarity to ideal solution method for ranking five alternative hotels. Future studies may extend the present model for various product selection problems for which product feature ratings are available.