Apriandy Angdresey, Indah Yessi Kairupan, Kenshin Geraldy Emor
Government agencies in this modern era are greatly assisted by the rapid development of information technology, so transparency and speed are imperative in providing services to the public. The use of social media by government agencies is one of the innovations by maximizing technology, one of which is the Ministry of Health of the Republic of Indonesia which is the ministry in charge of organizing government affairs in the health sector. The information contained on the Twitter social media account of the Ministry of Health of the Republic of Indonesia is various types of information uploaded randomly, therefore Twitter users often cannot distinguish the types of information provided by the Twitter of the Ministry of Health of the Republic of Indonesia. Based on this case, Twitter users can provide responses or comments that often lead to pros and cons. The extreme gradient boosting (XGBoost) algorithm is a tree-based algorithm such as the decision tree algorithm, which uses an ensemble principle that combines several weak learning sets and makes a new model that is strong to produce strong predictions. In this study, our aim is to implement extreme gradient boosting to classify the tweet topics and analyze the sentiment towards comments made by the public on the tweets of the Ministry of Health of the Republic of Indonesia. The results of this study indicate the highest level of accuracy in the classification, i.e. 89.35% with a precision of 88.76%, and 88.58% for the recall value from 2243 tweets. Furthermore, the best accuracy in sentiment analysis was obtained at 91.22%, 89.17% for precision, and for recall at 89.06% with 304 comment data. © 2022 IEEE.
Universitas Katolik de la Salle, Department of Informatics Engineering, Manado, Indonesia
