This study utilized integrated coherence to improve cryptocurrency market predictions. RoBERTa and DistilBERT models were employed for measuring integrated coherence and results were obtained by applying LSTM. The proposed DistilBERT and RoBERTa models were compared to normal coherence through the t-test. The results showed that both the proposed DistilBERT and RoBERTa models outperformed LSTM with normal coherence. The study's findings confirmed that user sentiment is a necessary factor for market predictions.
목차
Abstract Introduction Theoretical Background Long Short Term Memory (LSTM) RoBERTa 감성분석 모델 DistilBERT Methods Hypothesis Development Data and Method Process Result Conclusion References