Comparative Analysis of Polarity of Text-based Sentiment Analysis

Sentiment analysis deals with finding the polarity of the text. Polarity can be grouped into three categories: pessimistic, optimistic, or neutral. The general process for sentiment analysis follows four modules: the collection of data from various sources (social media, news, reviews, feedback, etc.), preprocessing of data (removing of features that do not produce any sentiment), applying approaches (machine learning, knowledge-based or hybrid) and produced output (polarity of text). Polarity detection is conducted on the document, sentence, or aspect levels. Analyzing opinion helps to understand the subjectivity of the data or to find whether the data is falsely generated. This study analyzed nearly 24 methods based on machine learning, deep learning, and other state-of-the-art techniques. Machine learning approaches such as SVM and NB and deep learning approaches such as CNN and LSTM have mainly used methods as they provide better accuracy than other models.