Penerapan Latent Dirichlet Allocation dalam Analisis Konten Memasak pada Youtube Indonesia

Youtube is a platform used to share videos. Since its founding in 2005, to date there are more than 20 million active users with 2 million videos uploaded every day. This is inseparable from the contribution of Indonesian content creators who also share videos with various types of content. One type of content that is loved by the Indonesian people is content about cooking or culinary belonging to food vloggers. In this study, the author wants to know what are the dominant topics related to the type of content created by food vloggers on their Youtube channel. This study uses the Latent Dirichlet Allocation (LDA) method. The research began by conducting text mining experiments on 3846 videos from the Youtube channel of nine food vloggers with more than one million subscribers. Then the LDA method is applied which is then analyzed to determine the optimal number of topics of content types by looking at the perplexity and topic coherence values. As a result, there are 5 topics of content types that often appear in videos belonging to food vloggers. Topics of this type of content include tips for making economical cakes, ingredients for making oven cakes, food business ideas, the process of cooking viral dishes, and easily available ingredients for making snacks.