Implementation of Bidirectional Long Short Term Memory (BiLSTM) Algorithm with Embedded Emoji Sentiment Analysis of Covid 19 Anxiety Level and Socio Economic Community

Abstract

The COVID-19 pandemic has presented multidimensional challenges in Indonesia, significantly affecting social, economic, and public health at the level of anxiety. Public anxiety related to the pandemic can be reflected in online media, especially Twitter, which is the main channel for information sharing and emotional expression. This study aims to understand the level of public anxiety in relation to the aftermath of the COVID-19 pandemic by using a classification method. Classification is carried out using the Knowledge Discovery in Database method with the Bidirectional LSTM algorithm and emoji embedding sentiment analysis, and K-Fold Cross Validation testing is also carried out with various optimizers. The final result of the best accuracy rate obtained was 98.08%. This shows that the classification model created is good.