Heather Bell Undergraduate Dissertation 2017/18

Improving Customer Satisfaction through Sentiment Analysis of Hospitality Reviews

Supervised by R.Gaizauskas

Abstract

The existence of online review sites has created a reality whereby potential customers of a product or service have access to hundreds of reviews, authored by fellow consumers. In the case of the hospitality industry, the opinions expressed within reviews are used by consumers to determine whether or not the establishment is worthy of their business. These reviews can also be used by hospitality establishments to gather feedback from past customers, identify issues, and analyse customer satisfaction. One of the biggest challenges for businesses in the hospitality industry is achieving customer satisfaction, so it is paramount that customer opinions are known to take action where necessary. However, the sheer volume of reviews makes manually analysing customer feedback a laborious and expensive task, where resources required to perform the analysis become strained. It is therefore necessary to apply tools that provide a method of identifying user opinions and enabling analysis of large volumes of reviews, in a way that reduces the strain on resources. The key to gathering these opinions within reviews is a branch of natural language processing, known as "sentiment analysis".

This project aims to implement a system, which successfully performs sentiment analysis on user reviews within the hospitality domain. The best ways to do so will be investigated, notably those that involve supervised machine learning. The success of the system's capabilities can be assessed by how informative the resultant output is, in terms of how easily it enables users (management of hospitality businesses) to identify aspects of their business that require improvement. Identifying areas of improvement, and acting on them, will enable management to achieve their goal of customer satisfaction.