Fine - Grained Sentiment Analysis on Online Reviews

Sentiment analysis is an effective text categorization technique that is often used. It analyses any given text or comment and categorizes it as positive or negative depending on the viewpoints expressed. Previous sentiment categorization research has relied on either a traditional-based approach or machine learning approaches. Similarly, a fundamental flaw in earlier systems was they focused solely on only two-way classification or tri-classification of reviews, i.e., classifying the review as positive, negative, or neutral. Denial of the highly positive and highly negative reviews will lead to a misconception of the user's opinion of a product or movie, causing a trend or business to suffer. When solely a lexicon-based method is used, the system is heavily reliant on the lexicon resource and dictionary chosen. The performance of a machine learning system is determined by the algorithms used. Beyond the binary or tri classification of the customer's review, the suggested work integrates a traditional based approach (SentiWordNet) with statistical methods such as Decision Tree (DT), Naive Bayes (NB), Logistic Regression (LR), and SVM are used to solve the aforesaid challenges. These machine learning methods were compared to a lexicon-based strategy in terms of performance. The results showed that the Logistic Regression and SVM methods outperformed the rest two algorithms.