Sentiment Analysis of South African News Company

Abstract

This paper reports on using RStudio packages to conduct a text and sentiment analysis of tweets from SABCNews. To download selected tweets, the twitteR package in R was used. A 10-day period was to collect data (tweets). In this regard, a set of 1000 tweets written in English were analysed from the @SABCNews. This analysis was performed to classify each tweet as positive, negative or neutral and to what degree; thus, learn more about the audience, transform every interaction into an actionable strategy for success, and shape the company’s business strategy. The tweeting pattern is measured using Hashtag and Favorite Count, Source of Tweet, and Count of Tweet compared to Favorite Count and Retweet Count. Based on trust, joy, surprise, anticipation, fear, sadness, disgust, and anger, it can be observed that most tweets expressed trust and, joy; with no tweet expressing anger. Furthermore, the Twitter sentiment analysis assigned for each tweet showed that a rapid change in sentiments for each tweet occurs late at night before individuals sleep and early morning before they go to their daily duties. While these results are positive for SABCNews; however, they emanate from only English language tweets. Therefore, future work will focus on sentiments expressed in other South African languages and compare views from other news agencies.