Web Services for Sentiment Analysis on Social Media Websites

Abstract

Since there are so many types of services available for web, it is hard to determine a service that will work best to analyze social media websites for the extremist content. An approach is developed for testing each dataset that is extracted according to the keyword "extremism" by working with APIs of web services utilizing PHP programming language. In making the comparison for web services, recall, precision, accuracy, and f-measures are analyzed to determine the service with the lowest mean square error (MSE). Four sentiment analysis web services are being used: Aylien, Sentiment Analyzer, Monkey Learn, and Parallel Dots. Monkey Learn scored lowest on MSE of 14% among all web services in our comparison of all web services.