TelcoSentiment: Sentiment Analysis on Mobile Telecommunication Services

Abstract

The mobile telecommunication services market in Malaysia has massively grown in the last decade. The competition among the existing and new telecommunication service companies became more intense as they are trying to maintain their existing customers and get new customers at the same time by making a lot of eye-catching promotions and seasonal events. These strategies are often compared by their customers on media social such as Twitter and Facebook. However, it will take a lot of time and effort to capture and review millions of tweets. This study aims to analyse the customer feedback and review on mobile telecommunication services in Malaysia using sentiment analysis approach. We develop a mobile application that analyse user reviews on mobile telecommunication services in Twitter data from Malaysia. The mobile application extracted data from Twitter, preprocessing the tweets in real-time, and visualise the sentiment analysis results using a bar chart, stacked area chart, and word cloud. We found that Naïve Bayes algorithm obtained the highest accuracy with 75.79% for ratio 9:1 and had the most informative features than other algorithms. We assessed the mobile application features using a preliminary evaluation. Many respondents agreed the mobile application is useful in visualising the sentiment analysis results and easy to use. The TelcoSentiment application can be used to suggest the public users in selecting the most preferred mobile telecommunication services. Furthermore, the application will be able to support telecommunication services companies to improve their service performance in the future.