Sentiment Analysis of Movies Based on Natural Language Processing

Abstract

Movies are an important spiritual food for people to satisfy their spiritual needs in contemporary society. Each person's evaluation of one movie is different. The data mining of movie-related data can make a certain accuracy prediction of the final rating of a movie. This paper shows a new movie rating prediction model is proposed by using nlp technology. It analyzes the data recorded in Movielens for nearly 45,000 movies. Some of the selected data include final rating, movie duration, title, budget, story overview, genre, and so on. This includes a large amount of textual data that needs to be converted into vector data using nlp techniques. In this paper, random forest and neural network are chosen as the main models and a lot of tuning is done to obtain a relatively accurate model, i.e., a random forest movie rating prediction model using natural language processing. This research can be used for upcoming or newly released movies that lack ratings, which is a good reference for audiences' viewing choices and movie producers' promotion and investment decisions.