The goal of product reviews is to identify any positive or negative undertones in a given text document. It is widely utilized in a variety of commercial fields to enhance goods or services by learning what customers think about them. Deep learning produces cutting-edge findings in a variety of difficult disciplines. Due to deep learning's success, several researches have presented deep-learning-based sentiment classification models, which outperformed traditional machine learning models. The challenge of modeling and predicting the helpfulness of online reviews look at the variables that affect review helpfulness and attempts to forecast it with accuracy. In this research, It is compared several deep learning-based sentiment classification model topologies to draw useful predictions for the analysis of sentiment based customer reviews. Also, it is provided that a benchmark comparison of many deep learning models such as the Recurrent Neural Network, Attention Mechanism, and the Bi-Directional Long Short Term Memory and evaluated using several word embedding techniques such as FastText and Word2Vec. Each model was evaluated using one of two different setups. The performance of algorithms are compared and suggested for the better use real time applications.

Deep Learning Methods, Sentiment Analysis, Online Review Data, Bidirectional RNN Model, Bi-Directional Long Short Term Memory.