Abstract:
Sentimental Analysis is a Natural Language Processing (NLP) procedure used
to characterize records for recognizing positive or negative reviews. Customer
happiness has recently emerged as one of the most important criteria in the restaurant
industry's success. The importance of consumer feedback cannot be overstated. For the
sake of social media, people are more motivated to read reviews before coming to a
restaurant. Customers who want to choose a restaurant may read a lot of reviews to get
a good idea of the restaurant's quality or services. As a result, a nostalgic classification
of a large number of audits is required to achieve meaningful experiences, allowing
customers to select a restaurant based on their preferences. Sentimental analysis can
help with this classification. This research suggests a system for categorizing customer
reviews into good and negative categories based on sentimental input. 1000 restaurant
evaluations from Tripadvisor, foodpanda, foodbank, and other restaurant review sites
were used to test the proposed solutions. In this paper Split Test, 20% Data Testing and
80% Data Training have been used. More specifically, the proposed system has been
tested with four supervised algorithms of machine learning: Support Vector Machine
(SVM), Multinomial Naïve Bayes, Random Forest and Decision Tree for sentiment
classification of comments. The untried result shows that this proposed system can
classify restaurant reviews with 71.50% accuracy using SVM, 73% accuracy using
Multinomial Naive Bayes, 70.50% accuracy using Random Forest, 65%using Decision
Trees.
