Aspect based sentiment analysis

Abstract

Nowadays, reviews have become one of the primary sources of feedback on many contents like the hotel, food, products, and more. These reviews are beneficial for the consumers and producers. We use sentiment analysis to analyze this review, which helps to give the polarity, i.e. positive, negative or neutral and allows producers to enhance their productions or businesses according to customer satisfaction, whereas consumers predict the future convenience. Most decision-making procedures are based on the decision maker's preferences and public perceptions of potential alternatives. In the multi-criteria decision-making field, user preferences have been heavily considered. Sentiment analysis, on the other hand, is a branch of natural language processing devoted to the creation of systems capable of analyzing reviews and determining their polarity. In this work, aspect-based sentiment analysis extracts the coarse-grained sentiments behind the hotel reviews depending on the specific aspects. In analyzing the sentiments of the aspects from the review, we first preprocessed the reviews and then feature selection was done. Then classifications of the sentiment polarity of the aspects were done using supervised learning models and deep learning models and various metric measures like F1-score, accuracy, precision and recall have been noted down.