Sentiment Analysis with Novel GRU based Deep Learning Networks
Chary Vielma, Abhishek Verma, Doina Bein · 2023
Consumers make decisions online on diverse websites based on recommendations from people they have never met. This shared online input provides thus insight into the way they perceive things such as products, services, or events, can be beneficial to some degree to the users who are looking to make decisions. Sentiment identification is an ongoing research topic that has successfully shown results for movie and product reviews, blogs, social media post. As users, we shop for the best fit. Text classification, a subcategory of natural language processing, is the task of categorizing text to represent the word and its use in context. Neural networks have been widely used to implement text classification mechanisms for sentiment analysis. In this research we aim to use movie reviews from the internet movie database (IMDb) dataset [1] to perform neural network sentiment analysis. Two novel multi-branch models are explored: CNN-GRU and CNN-bidirectional GRU. The results show that while CNN-bidirectional GRU has slightly higher accuracy, the CNN-GRU has a comparable accuracy and did so with less training time.