Topic Modelling of Merdeka Belajar Kampus Merdeka Policy Using Latent Dirichlet Allocation

Topic modeling is the process of representing the topics discussed in text documents. In the current era of internet technology development, digital data is growing increasingly large, including tweet data from Twitter. This research aims to obtain topic modeling related to the Merdeka Belajar Kampus Merdeka policy on Twitter, which has been classified into positive and negative sentiments. The topic modeling method used is Latent Dirichlet Allocation (LDA). This method is for summarizing, clustering, connecting, or processing data from a list of topics. The data used in this research are tweets with the keyword "Kampus Merdeka" uploaded on Twitter. A total of 1579 tweets with these keywords were classified into 648 tweets and 931 tweets, respectively, with positive and negative sentiments. Each tweet with positive and negative sentiment produces 5 topics with parameter values α and β of 0.1. The coherence value in topic modeling for tweets with a positive sentiment (0.44) is more significant than for tweets with a negative sentiment (0.38) and represent for drawing conclusions about topics based on relationship between keywords in negative sentiment is more challenging compared to those in positive sentiment to the Merdeka Belajar Kampus Merdeka policy on Twitter.