Shapley Additive Explanations for Text Classification and Sentiment Analysis of Internet Movie Database

Abstract

The application of Artificial Intelligence (AI) is increasing in areas like sentiment analysis and natural language processing (NLP). Automatic sentiment analysis provides a guide to capture the user emotions and classify the reviews into positive or negative. One of the challenges of using general lexicon analysis is its insensitivity to all domains. There arises a need for the interpretability of the output predicted from the AI sentiment analysis models. This paper developed a Shapley Additive Explanations for Text Classification (SHAP) based model to classify the user opinion texts into negative or positive labels. Our sentiment analysis model is evaluated on the Internet Movie Database (IMDB) datasets which have rich vocabulary and coherence of the textual data. Results showed that the model predicted 89% of the user reviews correctly. This model is very flexible for extending it to the unlabeled data.