Sentiment classification for review rating prediction based on hybrid reaction diffusion support vector neural network

Online reviews offer precious insight to business areas to improve the services, and products by the experience of the customers. The reviews contain the consumer's feedback regarding the usage of the product. Moreover, online reviews play a crucial function in manipulating customer’s purchasing decisions on social media. With the rise of e-commerce, people have an excellent opportunity to share their experiences on review websites. Due to the rapid increase in product evaluations, predicting review ratings has become more crucial in numerous applications. In this work, the hybrid reaction diffusion support vector neural network (RDSVNN) is developed for predicting the review rating. The input product reviews are collected from the Amazon review dataset. Feature extraction is the necessary step to extract the relevant features. The sentiment analysis is carried out utilizing the hybrid support vector machine (SVM) fused Rider optimization algorithm-based neural network (RideNN) called SVM-RideNN. The review rating prediction is performed using the RDSVNN. Furthermore, the metrics like recall, precision, recall-oriented understudy for gisting evaluation, and F1-score are employed to estimate the RDSVNN-based review rating prediction, in which the optimal values of 96.92%, 93.84%, 88.76% and 95.32% are achieved.