Sentiment Analysis on Online Learning During the Covid-19 Pandemic Based on Opinions on Twitter using KNN Method

Abstract

Coronavirus Disease of 2019 began in Wuhan in December 2019 and it was declared as a global pandemic by WHO. Until January 2021, it affected all of human activities on earth i.e., experiencing many obstacles from restrictions on activities, closure of tourist attractions to restrictions on face-to-face learning activities in schools or universities. Due to the policy of providing a broad influence on the community with various comments through social media, many twitter users make tweets containing positive and negative comments leading to statements about online learning or daring. The problem is that they contain so many different words, abbreviations, informal language, and symbols, creating difficulties to choose which words or groups of words that can produce positive or negative statements. K-Nearest Neighbors algorithm is used to classify positive and negative tweet data, the results were AUC for class 0: 0.754, 1: 0.635, 2: 0.721 and with a precision classification score of 0.86, recall is 0.85 so that the results of the classification of negative and positive sentences on the online learning tweet data were ROC-AUC of 0. 853 and the accuracy value of 0.885.