Blend of VADER and TextBlob for Movie Script Classification

This paper compares two sentiment analysis models: VADER (Valence Aware Dictionary and sEntiment Reasoner) and TextBlob on the classification of movie scripts. The scripts were rated as G, PG, PG-13, R, NC-17 which are standard MPAA ratings. Further, the work also extended the use of the blended score that combined the weighted results from VADER and TextBlob to reach higher accuracy, using certain features such as profanity scores. It talks about the working mechanism of VADER and TextBlob in application to the English movie scripts. The paper critically reviews the literature of methodologies pertaining to sentiment analysis, presents some practical experiments that establish the efficiency of our approach, and concludes with the correlation analysis of the sentiment scores between VADER and TextBlob while suggesting further research directions for multilingual sentiment analysis. While BERT and RoBERTa are really accurate, VADER and TextBlob have their merits in simplicity and computational efficiency, offering speed whereby prototyping is possible; hence, these would be better positioned for use against smaller datasets and resource-poor environments. Coupled with their interpretability, features like these make them quite good for movie script analyses, hence being able to get suitable insight into sentiment without heaviness due to higher-order models. A balancing act between the two sides will make the tools apt for this study: VADER and TextBlob.