Deep Learning for Aspect-Based Sentiment Analysis on Indonesian Hotels Reviews

Tourism is one of the fastest-growing industries. Many travelers book hotels and share their experiences using travel e-commerce sites. To improve quality products services, we can take advantage by analyzing reviews. We see good bad thing reviews in every aspect hotel. However, research to analyze sentiment Indonesian hotel still relatively new. In this work, propose create an Aspect-based Sentiment Analysis (ABSA) solve problem. This consists four steps: collecting data, preprocessing, classification, classification. Our classification process compares with eight deep learning methods (RNN, LSTM, GRU, BiLSTM, Attention CNN, CNN-LSTM, CNN-BiLSTM). have six classes aspects which are harga (price), hotel, kamar (room), lokasi (location), pelayanan (service), restoran (restaurant). analysis, compared two scenarios classify sentiments as positive or negative. The first all aspects, second aspect. results showed that LSTM achieved best model for accuracy value 0.926. For our experiments a better result than aspects. CNN gets average score 0.904.