A BERT-Based Technique on IMDb For False Movie Review Detection

Abstract

Social media's popularity has grown alarmingly over the past ten years. In comparison to prior times, everyone is using technology more frequently. Many businesses use data from multiple websites to generate relevant information that can then be used for business objectives. Nowadays, people blindly trust the reviews available online and form a perception of any movie even before watching it. On websites like Amazon, IMDb, and Rotten Tomatoes, there is a wealth of textual information about movies. IMDb movie reviews are used to forecast user ratings for the films. Researchers in the field of machine learning have investigated a number of approaches to implementing the procedure with the highest level of accuracy. This paper focuses on finding false IMDb movie reviews by using a deep learning technique called BERT. In the proposed work, the BERT-base-uncased type is employed, which utilized pandas, torch, and transformer that produced an accuracy of 93% on the IMDb dataset.