Opinion classification for IMDb review based using naive bayes method

Personal evaluations of each film are what make up film reviews. As a consequence of this, moviegoers will have a difficult time determining whether or not the theater lives up to their expectations. Analysis of people’s feelings is the most effective remedy for this problem. The process of assigning emotional labels to texts in order to determine whether or not they include positive or negative sentiments is referred to as "sentiment analysis." The Naive Bayes method was selected because of its capacity to classify information based on a calculation of the probability that each class has against the items in the data sample that was supplied. The best model was created using unlemmatized data, 600-size vectors, and Naive Bayes classification. It was able to get an f1-score of 77.66% and an accuracy of 77.81%. The system’s ability to anticipate positive and negative views is sensitive to changes in the vector’s magnitude. Results for both accuracy and recall suffer when using a vector size of 400 instead of 600. The disparity in recall and precision numbers demonstrates this.