Sentiment Analysis on Amazon Dataset using Transfer learning

Abstract

In this paper, we have proposed several Machine Learning, Deep Learning and Transfer Learning model which is totally based upon new cutting-edge technology i.e., Sentiment analysis. With the continuous growth in the technology field Artificial Intelligence is on its way to change the motive of technology. So, this project is not only applicable for the above-mentioned dataset but it’s applicable for every dataset which consists of customer review. We all know that after the corona wave every market is switching into Online mode. There are a numerous number of website or company that are using customer sentiment technique to get the insight about their thought for particular product. In our project we use Classification technique to identify whether the sentiment for the particular product is positive or negative. So, based upon the existing dataset we follow the proper NLP pipeline so that we can extract all the features and get a clean dataset. ML and DL has huge algorithm support so using different classification algorithm we apply the NLP vectorization technique. Above all we also apply transfer learning model that is built on the top of Deep Learning model, which enhance the accuracy of the model because it’s developed on pre-built model. The ultimate goal for a developer is to get best accuracy, precision, recall and F1 Score. So, judging on the above-mentioned parameter and on behalf of industrial sight we compare different model and try to prove which model is the best model for this particular task.