Sentiment and Context-Aware Hybrid DNN With Attention for Text Sentiment Classification

A massive volume of unstructured data in the form comments, opinions, and other sorts is generated real-time with growth web 2.0. Due to nature data, building an accurate predictive model for sentiment analysis remains a challenging task. While various DNN architectures have been applied encouraging results, they suffer from high dimensional feature space consider features equally. State-of-the-art methods cannot properly leverage semantic knowledge extract meaningful relevant contextual features.This paper proposes context-aware hybrid attention mechanism that {intelligently learns highlights salient context text. We first use integrated wide coverage lexicons identify text then bidirectional encoder representation transformers produce sentiment-enhanced word embeddings extraction. After that, proposed approach adapts BiLSTM capture both order/contextual information long-dependency relation sequence. Our also employs assign weight give greater significance Finally, CNN utilized reduce dimensionality local key analysis. The effectiveness evaluated on real-world benchmark datasets demonstrating significantly improves accuracy existing classification.