Skripsi
ANALISIS PENINGKATAN AKURASI METODE DISTILBERT DALAM MENGKLASIFIKASI TWEET MENGENAI COVID-19
Sentiment analysis is a fundamental task in Natural Language Processing (NLP). Social media is designed to enable people to share content quickly through electronic tools. People can openly express their minded on social media sites like Twitter, which later can be shared with others. During the recent COVID-19 outbreak, public opinion analytics provided useful information for determining the best public health response. In this study, researchers will improve BERT accuracy by using the DistilBERT method. The DistilBERT classification method is designed to reduce the size and increase the training speed of the two way encoder representation of the transformer model (BERT). The experimental results using the BERT method generate an accuracy value of 87%, while using the DistilBERT method increased the accuracy value by 10%, so that the accuracy value using the DistilBERT method becomes 97%.
Inventory Code | Barcode | Call Number | Location | Status |
---|---|---|---|---|
2307001501 | T93104 | T931042023 | Central Library (Referens) | Available |
No other version available