Sentiment Analysis on Hotel Reviews to Quantify Amenities

Abstract

In today’s tech-savvy time, online reviews of a particular product/service play a significant role in determining the decision of a potential buyer. These reviews are crucial not only for the buyers/customers, but also for the sellers as these reviews can have a huge impact on their revenue. Thus, it becomes very essential to analyze, harness and process these online reviews for gaining useful insights. In this paper, the broad aim is to summarize the reviews of hotels using sentiment analysis but the idea can be extended to any kind of product or service. The contribution of the study is to develop a model/prototype to quantify the features/amenities (words that describe the hotel) of various hotels as a percentage of their positive and negative rating given by the user in their review. From quantified scores obtained by the prototype, further a complete summary about the product can be generated. Although, work has been done in the direction which accepts reviews from users, summarize the reviews and provide qualitative analysis of the reviews, but there had been very few works to quantify the features of a product based on their opinion words.  This quantitative approach impacts the human decision quality to a very large extent. This work has manifold benefits which can be explained as follows; at the user end it provides the user a complete amenity based quantified summary of the product, i.e. the user gets a concrete idea about each and every attribute of the product under consideration. For instance, let us consider a hotel as a product. The user gets a summary incorporating amenities of the hotel like rooms, staff, location, terrain etc. On the other hand, the seller of a product or service gets the idea about what amenities he can put forth as the product’s unique selling proposition and what amenities of the product/service needs improvement.