Sentiment Analysis of Peduli Lindungi Application Using the Naive Bayes Method

Abstract

Peduli Lindung application as a form of government policy in the context of handling Covid-19. The level of usability in an application is really needed to see the usefulness of the application itself. The analysis is carried out in the form of a fine-grained sentiment analysis based on a five-star review. Models used in conducting the analysis in this study using Naïve Bayes. Data used in get it through Google Play Store until April 2022. Rating 1 has the most number from other ratings, namely as many as 467 reviews and rating 4 has the lowest number, namely 55 reviews. The data is classified as negative as many as 146 data, a lot of data are classified as negative classified as true positive as many as 30 data, and data classified as neutral as many as 30 data, with classification accuracy still at 73%. The results obtained by the community tend to show words that refer to the problems that exist in the application.