<![CDATA[Learning media innovation is currently required to keep up with the development of science and technology. Duolingo is a free language learning application. Duolinggo has been downloaded more than 500 million times and recorded more than 21 reviews in the comments column consisting of positive and negative comments. Duolinggo user reviews are classified into two sentiments, namely positive sentiment and negative sentiment. Sentiment analysis is an activity used to analyze a person's opinion or opinion on a topic, to support the classification, the algorithm used is DistilBERT. DistilBERT is a technique of how to make the BERT model smaller, but has similar qualities to a large model, distilBERT can be termed as 2 running models, namely the teacher model and the student model, the teacher model is a large model and is trained with a complete range of features such as the base (pre-trained model) The results of the DistilBERT algorithm for classifying 1000 reviews of the Duolingo learning application produce precision, recall, f1-score values on class 1 labels are 74%, 96%, and 84%, indicating that this BERT algorithm is very good at predicting label classes. With the accuracy result obtained is 80% in 85 seconds.]]>
Copyrights © 2024
