Sentiment Analysis Classification using Deep Learning Techniques

People are increasingly expressing their ideas on websites in this day and daily, enormous amounts of client feedback are produced. These unstructured text reviews and customer comments are frequently used by customers when making decisions. However, because there is so much consumer feedback, it is time-consuming to read them all. The ability to forecast the precise sentimental polarity of user textual feedback evaluations for particular entity is still a difficult task because of phrase length constraints, textual order variations, and logical complexities. Bidirectional Recurrent Neural Network based on Self Attention Mechanism (BRNNA) for subjectivity categorization of reviews was developed to overcome these issues. The described method uses pre-trained word embedding to minimize text representations, avoiding problems with data sparsity. Moreover, the attention mechanism uses multiple word and phrase weights to capture n-gram attributes and concentrate on the most important context-relevant information. The suggested model automatically learns classification features and captures the semantic and spatial information that is essential in detecting the polarities of sentiments utilizing Bidirectional Recurrent Neural Network (BRNN) and Self Attention Mechanism. The proposed BRNNA model is evaluated in comparison to various benchmark techniques. The accuracy of the proposed approach is 91.06%