Tools
Tools
Higgins, Liam (2022) Classification of Airline Customer Sentiment Expressed in Twitter Tweets using Lexicons, Decision Tree, and Naïve Bayes. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (822kB) | Preview |
Preview |
PDF (Configuration manual)
Download (700kB) | Preview |
Abstract
This paper describes Natural Language Processing (NLP) and Machine Learning approaches to Sentiment Analysis of Twitter Tweets relating to commercial passenger airlines. By extracting and analysing textual data obtained in real-time from the social media platform, Twitter, the research proposes a methodology to collect, process, and interpret emotional responses contained within Tweets. Two main approaches to classifying sentiment are described. Firstly, lexicon-based approaches using three valence lexicons (Syuzhet, Afinn, and Bing) and one emotion lexicon (NRC) to determine the semantic orientation of words found within Tweet text are discussed. Secondly, two supervised machine learning classification algorithms (Naïve Bayes and Decision Tree) are used to perform sentiment classification. The goal of the research is to provide a diverse and commercially useful method for airlines to monitor customer sentiment relating to their experiences of airline services. The importance and commercial application of obtaining customer insights from Tweets which have been posted online and describe customer experiences and attitudes is discussed. The paper aims to provide airlines with a means to improve service offerings, differentiate from competitors, and gain competitive advantage based on analysing customer sentiment to their services. A maximum accuracy of 71% was achieved using a Naïve Bayes classifier algorithm.
Actions (login required)
 |
View Item |