Analisis Sentimen Masyarakat Terhadap Konten Bahasa Isyarat Pada Media Sosial Tiktok Menggunakan RNN dan BERT

This research aims to explore public sentiment towards sign language content on Tiktok social media using a deep learning approach with RNN(LSTM) and BERT models. Data is collected, processed, and labeled to form a dataset used in analysis. The sentiment classification results from the RNN(LSTM) model show an accuracy level of 0.90%, indicating the effectiveness of the model in distinguishing sentiment between the two classes. Meanwhile, the BERT model achieved an accuracy level of 0.80%, showing the model's ability to classify sentiment from the data provided. Combining RNN and BERT models shows increased performance in classifying sentiment, that the combined approach can improve the model's understanding of complex content. In addition, the combination of the BERT model with RNN has also proven effective in overcoming the complexity of sentiment analysis in the context of sign language on social media. Continuous evaluation and validity of the model involving experts in sign language is recommended to ensure the sustainability and validity of the developed model. It is hoped that this research will make an important contribution to understanding the influence of social media on society's views of sign language and hearing disabilities in general, as well as strengthening efforts towards a more inclusive society.