An Automated Approach to Spoiler Detection and Elimination in Movie Reviews Using BERT NLP-Based Model

Abstract

This paper offers a browser extension that can warn against spoilers on review sites, and this is an attempt to deal with unintentional plot revelations. Using the data source from Kaggle, the data preprocessing techniques were used to enhance the data. The methodology consisted of training a spoiler detection model with BERT architecture in place, which can be assessed in terms of accuracy, precision, and recall as performance metrics. The experimental setup took place in a Kaggle Notebook environment and it also had a defined set of hardware and software specifications. The extension made it possible to do this through techniques of web scraping, etc. for instant spoiler detection. Discussing the accuracy issues of data and BERT token size limitations was mentioned. In general, this research presents the comprehensive solution of preventing spoilers to ensure the movie-watching experience in the digital era.