Topic modeling of disaster based on indonesia tweet using latent dirichlet allocation

Iswanto Iswanto, Alfian Ma’arif, Aninditya Anggari Nuryono


Twitter is one of the most used social media to spread information, which can be modeled using topic modeling techniques. In this study, disaster-related topics would be modeled. Several phases were performed to get the desired topics: data collection, stop word removal, stemming, case folding, normalizing slang words, and normalizing abbreviated words. The sampling of six topics was done as the LDA method parameter. Keywords used in data collection were taken from official Twitter accounts, such as The Meteorology, Climatology, and Geophysics Agency (BMKG) and the National Disaster Management Agency (BNPB). Keywords used were in Indonesian: banjir (flood), kecelakaan (accident), longsor (landslide), and others. Topics with high coherence values and the most discussed were coronaviruses, accidents, drowning, storms, windstorms, cyclones, and hurricanes. World Cloud was used to visualize the disaster-related topics.


Disaster; Indonesia; LDA; Topic modelling; Twitter



  • There are currently no refbacks.

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

shopify stats IJEECS visitor statistics