Opinion Mining: How efficient are Online classification tools?

Abstract

Recently, Online Social Networks (OSNs) are considered as important resource of information, since they provide a huge amount of data that reflects the interactions between users in various fields, such as: politics, sport and business. Opinion mining (or sentiment analysis) is a process that uses natural language processing, and text analysis methods to understand users’ feelings or opinions, and detect their polarity, which could be positive, negative or neutral. The outcomes of opinion mining approaches help in extracting useful patterns that enable traders to take critical decisions for business, marketing and politics. In the literature, we have several proposed opinion mining systems, tools and approaches, but in general they are not available on public. Many other online opinion mining tools are simple to use and available for free or as demos. Opinion mining online tools performance need to be evaluated to attract researchers and companies utilizing their advantages. The main purpose of this study is to evaluate how efficient are online opinion mining tools for Arabic language. We used benchmark Arabic opinion collections and classify them using two popular online sentiment analysis tools that support Arabic language; Paralleldots and Repustate. The experiment used prediction quality measurement to evaluate these tools and compare their results with several machine learning classifiers in order to recommend the best available solution for Arabic sentiment analysis. Our results showed that Paralleldots API is highly recommended for Arabic sentiment analysis for both positive and negative reviews. © 2020, World Academy of Research in Science and Engineering. All rights reserved.