Tools
Tools
Alaba, Adedoyin (2021) Detecting Spam Campaigns on Twitter Using Machine Learning Approach. Masters thesis, Dublin, National College of Ireland.
Preview |
PDF (Master of Science)
Download (643kB) | Preview |
Preview |
PDF (Configuration manual)
Download (32kB) | Preview |
Abstract
A secure voting system entirely depends on modern technology. Cybersecurity guards against election violence and manipulation. Spammers have taken advantage of the huge popularity of social media to distribute spam messages because it takes advantage of relationships between users. Social spamming is more effective than conventional techniques like email spamming. One of the most significant reasons is that social media assist in the development of intrinsic trust relationships between online buddies, even if they do not know each other in person. Detecting spam is the first and most important step in combating spam. This study is focused on Twitter, and proposes a novel, effective approach to detect and filter unwanted tweets, complementing earlier approaches in this direction. Previous studies rely on historical features of tweets that are often unavailable on Twitter after a short period of time, hence not suitable for real-time use. This study approached an optimized set of readily available features, independent of the historical textual features on Twitter. This paper focuses on identifying SPAM tweet patterns used during elections by assembling two machine learning algorithms and applying them on the combination of unigram and bigram words to produce a better accuracy of 99.2%.
Actions (login required)
 |
View Item |