Unsupervised Sentiment Analysis of Amazon Fine Food Reviews Using Fuzzy Logic

Sentiment analysis has been increased popularity among the social media. The sentiment classification with machine learning identifies the positive and negative sentiment but the problem is that it doesn’t identify the duplication of reviews by the same individual. In this study, the sentiment of food reviews is assessed by employing a set of fuzzy rules and this suggested fuzzy system uses NLP (Natural Language Processing) techniques accompanied by an experimental unsupervised fuzzy rule-based system consisting of nine rules to classify the reviews into positive and negative. Unsupervised fuzzy rule-based systems are particularly advantageous in handling intricate and disorganized data, where the interrelationships among variables may not be precisely defined or readily apparent. This research took the dataset from Amazon fine food reviews and explores the more insight of food reviews. The proposed method considers the unlabeled text that utilizes the VADER tool with Fuzzy based system that outperforms among the other state-of-art algorithms with 86.3% of accuracy.