Browsing by Author "Al Kashari, Zainab"

Now showing 1 - 1 of 1
  • Thumbnail Image
    Item
    Mining Dubai Government Twitter Accounts
    (The British University in Dubai (BUiD), 2019-11) Al Kashari, Zainab
    Social media plays a critical role in the public sector as it allows the government to interact with the citizens. With the United Arab Emirates being active on social media platforms, this study then aims to identify the level of citizen engagement in Dubai government’s Twitter through the use of the data mining techniques. Decision makers in the government entities aim to engage the citizens in the various activities to better understand their perceptions, needs and expectations and accordingly take better informed decisions. This supports the transparency and trust of the government decisions. In this dissertation, the purpose is to contribute to the current research on social media by filling the gap on how local governments, especially in the United Arab Emirates, can increase citizens’ engagement on Twitter as preferred social media channel. Post engagement the total number of citizens’ interactions with a tweet and can be measured using different tweet attributes including retweets, mentions, hashtags and likes among others. Moreover, this study investigates the impact of the twitter post characteristics on the citizens’ engagements level. Thus, we collected, prepared and processed 74,037 tweets that represents all tweets for Dubai government twitter accounts during 2018. These tasks were followed by statistical analyses of the impact of post characteristics on the citizens’ engagement level. Next, we implemented various machine learning models to evaluate the performance of using the post characteristics and post content to predict the engagement level of citizens. Results indicate that citizen engagement level in Dubai government’s Twitter is significantly impacted by all post characteristics. It is also revealed in the study that citizen engagement is higher during weekdays compared to weekends. Furthermore, the machine learning models achieved promising results to predict the citizens’ engagement with highest accuracy for Random Forest and Linear Support Vector Machine of 78.3% and 78.2% respectively.