SENTIMENT ANALYSIS : A FRAMEWORK BASED ON ENSEMBLE METHOD

This research work aims to develop a novel framework that considers the extracted topics from previous work and then predicts the sentiments of the reviews and establishes a positive and negative class through clustering by using Possibilistic Fuzzy C Means (PFCM) Algorithm. The research work is further improved through an effective ML classifier known as Selective Memory Architecture based on Convolutional Neural Network (SMA-CNN) for multiclass classification. Moreover the considered classification speeds up the training process by upholding the efficient accuracy.  The above work will be concluded by using experimental approach by comparing the existing work with the proposed methodology which showcases the performance through accuracy. Analysis will be carried out with ground truth dataset along with synthetic dataset that is mapped with real word stream of data to exhibit an application of the proposed method.