SENTIMENT CLASSIFICATION USING N-GRAM IDF AND AUTOMAED MACHINE LEARNING

Abstract

Sentiment classification is a crucial task in natural language processing (NLP) that involves identifying the emotional tone of textual data. With the increasing use of social media, online reviews, and digital communication, analyzing user sentiment has become essential for businesses and decisionmaking processes. Traditional methods often rely on manual feature engineering and predefined models, which can be timeconsuming and less effective. This project proposes a sentiment classification system using N-gram, Inverse Document Frequency (IDF), and Automated Machine Learning (AutoML) to improve accuracy and efficiency. The system utilizes N-gram techniques to extract contextual features from text by capturing sequences of words, while IDF helps in assigning importance to words based on their frequency across documents. These features are combined to form a robust representation of text data. Automated Machine Learning is then applied to automatically select the best model and optimize hyperparameters, reducing the need for manual intervention. The approach leverages various machine learning algorithms such as Logistic Regression, Naïve Bayes, and Support Vector Machines. Experimental results demonstrate that the proposed system achieves high accuracy in sentiment classification tasks, outperforming traditional methods. The integration of AutoML significantly improves model selection and performance. However, challenges such as handling sarcasm and contextual ambiguity remain. This system provides an efficient and scalable solution for sentiment analysis in real-world applications.