A Hybrid Model for Sentiment Analysis Based on Movie Review Datasets

Abstract

The classification of sentiments, often known as sentiment analysis, is now widely recognized as an open field of research. Over the past few years, a huge amount of study work has been carried out in these disciplines by utilizing a wide variety of research approaches. Due to the possibility that the performance of sentiment analysis may be impacted by the high-dimensional feature set, text mining demands careful consideration during in the construction and selection of features.The process of recognising and extracting subjective information from written data is referred to as sentiment analysis. Sentiment analysis enables companies to understand the social sentiment around their brand, product, or service by monitoring the conversations that take place in internet chat rooms. In order to categorise people's attitudes or sentiments, this study provides a hybrid model (Support Vector Machine, Convolutional Neural Network, and Long Short-Term Memory). The findings of using the network model to sentiment analysis on the movie review or amazon review datasets reveal that it is possible to gain a good classification impact by using the model. The preprocessing is used for text mining, the removal of punctuation, and the generation of vocabulary, also uses GLOVE for vectorization and TF-IDF algorithms for better feature extraction. The results that were proposed were compared with various base models such as KNN, and MNB, amongst others, which demonstrates that the hybrid model performs better than other models.