Analysis of Tweets about International Leaders

By K. Chen (Assignment 5 for JMSC 6116, 23 April 2019)

The tweets analyzed were posted over a seven-day period between 15 to 21 April 2019. During this time, a number of significant events happened to some of the leaders profiled.

On 15 April, a fire ravaged the Notre Dame Cathedral located in Paris, France. As shown on the dashboard, this event triggered a spike in tweets containing #EmmanuelMacron on 16 April. The sentiments of these tweets were neutral in the beginning, and turned positive the following day. The top words picked up in these tweets included “cathedral”, “fire” and “rebuild”. There are only 539 posts picked up as I have searched only for English-language tweets.

On 17 April, North Korea announced that it had tested a new type of tactical guided weapon. This resulted in a sharp increase in tweets containing #KimJongUn on 18 April, although the number of posts dropped quickly in the following days. It is interesting to note the #KimJongUn hashtag has attracted the highest proportion of verified accounts among those of other leaders. This may be because a lot of these English-language tweets are posted by news organizations and commentators. The high proportion of retweets, the high number of tweet followers (compared to the number of accounts that the tweeters follow) and the large number of tweets posted per week are further indications that those posting the #KimJongUn hashtag are mainly professional (as opposed to amateur) accounts.

On 18 April, a redacted version of the Mueller Report was released in the US. The report documents the findings of Special Prosecutor Robert Mueller regarding the alleged interference by Russia into the 2016 Elections. As a result, the number of tweets mentioning #realDonaldTrump and/or #VladimirPutin spiked between 18 and 19 April. The nature of these two groups of tweets is, however quite different. First, the vast majority (or 85 per cent) of tweets mentioning #realDonaldTrump are original ones, versus only 55 per cent for #VladimirPutin. Only 5 out of 857 posters of #realDonaldTrump are verified accounts, compared to 15 out of 282 for #VladimirPutin. Finally, those posting #realDonaldTrump put up only 20 status updates per week, compared to 79 by those mentioning #VladimirPutin. These indicators suggest that most of those mentioning #realDonaldTrump are regular Twitter users while those of #VladimirPutin are, similar to the case of #KimJongUn, largely professional accounts.

The other three world leaders in the group, namely Xi Jinping, Angela Merkel and Theresa May did not go through significant events during the week under review. Although #XiJinping posts spiked on 18 April, no significant reason can be found. In fact, when #XiJinping post numbers are compared to those of #KimJongUn (see chart below), one can see that the increase is not significant.

For the text analysis of the tweets, I have used the Bing dictionary. I cleaned the tweets by converting the text into all lower case, and removing punctuation marks, hastags, mentions, URLs and extra spaces. I have also applied the wordstem function to both the dictionary and the tweets. To identify the top words used, I have excluded stop words and single letters.

Finally, I have put up an extra draw-down menu to allow the user to select another hashtag for comparison, providing a “Nil” option if no comparison is required. By using the conditionalPanel function in the UI, I have programmed the dashboard to include an extra line in each of the Post History and Sentiment Score History plots if a second hashtag is selected. A second set of boxes and tables will be shown for comparison as well.

Related posts:

When you get to be my age, mid-thirties, there is an expectation about your life held firmly by your peers. Some of these expectations come from society’s beliefs about aging. And some of these…