What's Next?: Exploring Machine Learning-Based Approaches to Content Suggestions Using IMDb Movie Reviews

In today's data-driven world, people often rely on online reviews to decide what to watch next. Streaming services, such as Netflix and Disney+, apply machine learning approaches in recommending films based on viewing history. Sentiment analysis refers to discerning and classifying the sentiment or opinion expressed by users towards various services, such as movies, product concerns, or events. This analysis involves categorizing the sentiment as positive, negative, or neutral. This study aims to create a system that utilizes sentiment analysis to evaluate movie reviews obtained from IMDb to assist viewers in making informed decisions when selecting films that suit their preferences. Sentiment analysis (SA) simplifies the process of opinion summarization by extracting the sentiments the reviewer conveys. The algorithms employed for accuracy rate classification include the Logistic Regression model, Linear Support Vector Machines, and Multinomial Naive Bayes. The study yielded a precision rate of 0.91 from the analysis of 6,157 data evaluations, suggesting a substantial presence of negative resistance. A data splitting technique was employed, where 25% of the dataset comprising 12,500 reviews was allocated for training purposes. The level of accuracy attained was 90.288%.

I. Introduction

Complex sentiments and convictions, such as demeanor, emotion, evaluation, or perspective, substantially impact human behavior [1]. The extent to which others' perceptions of the world influence our decisions largely depends on the fact that conveying others' opinions is hardwired into every individual and defines us as “social creatures.”