Track: COVID-19 Analytics Competition

Abstract

The increase in positive cases of Covid-19 in Indonesia is still relatively high. The main factor affecting the high number of positive cases of Covid-19 in Indonesia is the decreasing level of citizen compliance. This study aims to build a model of citizen compliance representation through a sentiment analysis approach using social media. The method used to build this model is done with a text mining approach, using the sentiment analysis by Support Vector Machine (SVM) algorithm. This model was built using the term Large-Scale Social Restrictions (LSSR) as a key phrase. This key phrase is interpreted through four main types of activities, namely work from home (WFH), studying at home, transportation, and physical distancing. The data used is data on twitter for the period April-August 2020. This representative model of citizen compliance during the Covid-19 pandemic is able to show a visualization of the condition of citizens' compliance levels grouped by city and province. The citizen compliance representation model shows an accuracy rate of 98% for WFH activities, 93% for learning activities at home, 88% for transportation activities and 90% for physical distancing activities. This model can be an initial reference for identifying the level of compliance of citizens with regard to government policies related to programs for handling Covid-19 cases in Indonesia.