USE OF TWITTER DATA FOR DISASTER DAMAGE ASSESSMENT IN BANTEN AND LAMPUNG, INDONESIA, BY ANAK KRAKATAU TSUNAMI EVENT IN DECEMBER 2018

• Main Author: Widiyanto, Wirawan
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/43753
• Institution: Institut Teknologi Bandung

Indonesia is a disaster-prone country, with the last major disaster is Anak Krakatau Tsunami that happened on December 22ND, 2018. Disaster management is all activities involving organization, planning, and application of measures preparing for, responding to and recovering. The Ministry of Public Works and Housing establish Disaster Management Task Force to strengthen disaster management within the Ministry. One of the tasks of the Disaster Management Task Force is to produce disaster damage assessment but restricted by the lack of budget, personnel, and logistic. Social media data especially Twitter data can be an alternative to obtained data for disaster damage assessment. To use Twitter data to produce damage assessment information constrained by the nature of Twitter data. Therefore, it is essential to find a mechanism that can automatically extract infrastructure damage information from Twitter data. The objective of this study is to build a dictionary that can help to sort infrastructure damage information from twitter data. In order to build infrastructure damage dictionary, we also need to find a term that related to damage and infrastructure. At first, this study used unigrams tokenization from language model to find what is the word related to damage and infrastructure that people used in their tweets concerning with Anak Krakatau tsunami event. The next step, we used bigrams tokenization from language model to produce paired_words dictionary. From 52936 tweets, can be filtered into 19 tweets that mention the damage on infrastructure. Most of the tweets that express information on infrastructure damage come from the news. Only two tweets that express infrastructure damage because of Anak Krakatau event but it does not mention the exact location or how big is the impact. From these finding, we can conclude that Twitter data constructed in the event of a disaster can be used as damage assessment because it contains information on infrastructure damage.