Sentiment Analysis against the Community Activities Restrictions Enforcement (PPKM) Data using ANN (Artificial Neural Networks)

Abstract

Enforcement of Community Activities Restrictions Enforcement (PPKM in Indonesian) was announced and enforced in July 2021 and has had a significant impact on many communities in Jakarta, with social networking media is becoming popular as a powerful tool for influencing people and sharing their views with the general public, to understand the emotions of the public in social media sentiment analysis is carried out using deep learning (ANN) with tweets dataset related to the enforcement of community activity restrictions (PPKM) in Indonesia. The results of this study show that while the model is being trained, the model tends to overfit after achieving the highest accuracy at a particular epoch. Comparing the three experiments, using features with TFIDF values above the median makes a small contribution to improving the model, while using ngram doubles the number of important vocabularies and other vocabularies. An important vocabulary for producing more influential features that can be combined with, which can significantly improve the accuracy of the model.