Arabic blogging Sentiment Analysis
, Lama Alsudias,Mostafa Salah, Fathy Essia . 2014
Today, microblogging has become the most popular communication tool among users of social networks. Many users share their opinions on different fields, and those users speak different languages, are of different ages and education levels, and the like. Consequently, opinion mining has become an interesting area of research. Arabic is one of the most used languages in the world. In this paper, we build a machine learning-based sentiment analysis system for mining and analyzing the Arabic tweets in social networks to determine positive and negative sentiments. We also built an application that determines the percentage of positive and negative opinions based on certain hashtags in specific domains. Regarding sentiment mining and analysis, many points will be addressed: building an Arabic corpus for tweets, filtration of the tweets’ tokens, and building a fault tolerance-based classifier. The classification uses a fault tolerance technique and different machine-learning algorithms (Support Vector Machine (SVM), Naïve Bayes (NB), and Decision Tree (DT)). A prototype was built and evaluated. The study yielded the following results: the average accuracy of the work based on the voter model is 84.5%. The application runs on the user’s selection of the hashtag name and its domain. The areas assessed are educational, social, economic, athletic, and political. It shows the percentage of positive and negative tweets in addition to the number of tweets written in this hashtag and the time it takes to process a calculation.
