Event Detection News Content on Twitter Using DBSCAN Algorithm

Authors

  • Muhammad Fakhri Sadewo, Gifari Muhamad Qudsi, Angga Agustian, Raka Haidir Rakhman, Sukenda, S.T., M.T.

Abstract

Twitter is a social media and microblogging service that allows users to send realtime messages. This message is popularly known as a tweet. Tweet is a short message with a length of characters limited to 140 characters. Due to the limited character that can be written, a tweet often contains abbreviations, slang language or spelling errors (Agarwal et al., 2014). Since its inception, Twitter was created as a mobile-based service designed according to the character limits of a text message (SMS), and to this day, Twitter can still be used on any cell phone that has the ability to send and receive text messages (The Twitter Government and Election Team, 2014). Indonesia is the third largest tweet-producing country with six million tweets per day [2]. This can be a huge potential for information to be exploited. For example, to find the latest news information that is currently happening in Indonesia. There is a lot of information on Twitter that is up-to-date and of course very useful for some people. However, there were also tweets that were completely uninteresting to some people. It takes a way to determine reliable information in the event of an incident on Twitter. After doing research on twitter data with the keywords "corona virus in Indonesia" on 20,000 tweet data, it can be concluded with several stages of the process such as the preprocessing process. And at the DBSCAN stage, the results obtained are in the form of 0.1% noise belonging to cluster 0 and 99% normal which is included in cluster 1 with an epsilon value of 106.65298405657 and a min point value of 999.

 

Published

2020-12-10

Issue

Section

Articles