Arabic Tweets Sentiment Analysis using Hybrid Approaches
Manal Essam, Mohamed R. Elmenshawy, Hamdy M. Mousa · International Journal of Computer Applications · 2020
The unrestrained-access to social media makes people share their daily life; a Twitter platform allows its users to openly express and share their emotions about several issues in a predefined length.Thus, it becomes one of the most dominant networks in Arabic countries.Therefore, the Sentiment Analysis of Arabic tweets is a practical task of analyzing common sentiments and feelings.However, the existing resources regularly focus on the English language due to the shortage of Arabic Sentiment resources.In this paper, a new sentence-based sentiment analysis system had developed for Arabic tweets.Initially, the main sentiment classification approaches had applied for the sentence-level to obtain the most suitable one.As a result, the steps towards the construction of a new dataset had evaluated.The experiments show that the supervised approach is the most accurate one, especially with the absence of the Arabic dialects' (informal) lexicons.Experiments comparisons achieve satisfactory results with high accuracy (78.08%) by supervised approach, Unsupervised gives acceptable accuracy (75%), and Fmeasure (74.1%) using a Hybrid classifier.