Penggunaan Recurrent Neural Network Dalam Mendeteksi Sentimen Berbahaya Pada Platform Media Sosial

Chuanta, Roy Vidia
Harahap, Mawaddah
Putra, Adya Zizwan

JURNAL TEKNOLOGI DAN ILMU KOMPUTER PRIMA (JUTIKOMP)
Vol. 7 No. 1 (2024): Jurnal Teknologi dan Ilmu Komputer Prima (JUTIKOMP)

Social media platforms in today's modern era make it easier for people to communicate and socialize. Behind that, on a hot topic, there must be sentiment. Every sentiment the community conveys varies; some are good and neutral, and some are bad or dangerous. To detect sentiment, start from crawling data, pre-processing, labeling, and then testing or training to get accuracy value, recall value, f1 value, and precision value using Long Short Term Memory. They obtained an accuracy value of 0.582, recall value of 0.582, f1 value of 0.428, and precision of 0.339. This LSTM model can be used to develop an analysis model that is successfully achieved.