Analisis Sentimen Data Twitter Topik Politik Dengan Metode Naive Bayes Dan Convolutional Neural Networks (Cnn)

Abstract

Sentiment analysis is a method used to identify, derive, and evaluate sentiments, opinions, or emotions contained in text. In the context of political topics, sentiment analysis can help understand public views and attitudes towards issues, presidential elections, presidential candidates, parties, etc. This information is valuable for decision makers, including the government, in the image of political figures and political life in Indonesia. This information is invaluable for decision makers, including the government, in the image of political figures and political life in Indonesia-data retrieval, data labeling, data pre-processing, data extraction, CNN classification, and naive bayes This shows that the accuracy value of the label between the negative and the requested information gives a greater true negative value than other sentiments. The large percentage recall value found in Naive Bayes with negative sentiment proves that, the ability of the model the number of correct ones is 0.95. Meanwhile, the highest percentage value of f1-score is obtained in Naive Bayes with a value of 0.81. Naive Bayes obtained an accuracy of 0.69 while CNN obtained an accuracy of 0.68. The above results show that there is a fairly thin difference in the accuracy obtained. So it can be concluded that the two models above for twitter sentiment analysis on political topics are good enough to use the above methods.