Sentiment Analysis Technique for Textual Reviews Using Neutrosophic Set Theory in the Multi-Criteria Decision-Making System

In recent times, numerous decision-making procedures are not only based on the decision-making of choices, but also public perceptions of possible solutions. In a multi-criteria-based decision-making system, user preferences have been deeply considered. Sentiment analysis, on either side, is similar to natural language processing dedicated to the creation of methods capable of assessing evaluations and determining their intensity. The main aim of this research is to make efficient decisions using social media tweets. The proposed method uses the SentiRank method and neutrosophic set theory to make decisions and rank the reviews. Novel multi-criteria-based neutrosophic theory is used in this research for decision-making. An assembled neutral vocabulary, and the adapted VADER, are used to create Neutro-VADER, a novel version. Every evaluation of a product feature is given a positive, neutral, or negative scores of sentiment by the Neutro-VADER. A unique idea at this level is to use the positive, neutral, and negative scores on emotion to represent reality, uncertainty, and falsehood participation levels of a neutrosophic number. The testing findings support the value of sentiment data through reviews in the ranking procedure. The performance metrics used in the systems are precision, recall, and F1 measures and accuracy for evaluating the aspect detection module. The system performs better in food, service, and pricing categories, whereas the anecdotes group gives bad results. F1 and accuracy level shows better results in the proposed system by using SentiRank and the neutrosophic set theory method.