Malaysian fibre internet service provider: a naïve Bayes classification Twitter sentiment analysis

Abstract

In the highly competitive landscape of Malaysian internet service providers (ISPs), users seek efficient ways to assess service quality. While various websites allow visual comparisons of fiber ISPs, a direct side-by-side evaluation remains elusive. A survey of 101 respondents revealed that 92.1% found researching a company’s reputation time-consuming. Additionally, relying on English-centric online ratings may lead to skewed outcomes, disregarding reviews in diverse languages. In response, we developed a web-based dashboard utilizing Twitter sentiment analysis (SA) and the naïve Bayes (NB) algorithm to classify Malaysia’s best fiber ISPs. The SA focused on four key factors: package price, internet speed, coverage area, and customer service, simplifying the comparison process. The system’s usability and functionality tests showed that both the English and Malay models could classify scraped Twitter data with an accuracy of 80%. The system’s remarkable usability score of 94.58% on the system usability scale (SUS) confirms its acceptability and excellent performance in achieving research goals.