Arabic Sentiment Analysis on Mobile Applications Using Levenshtein Distance Algorithm and Naive Bayes

Due to the specificity of the characteristics and features of the Arabic language and its complexity, the field of sentiment analysis in Arabic texts still represents a major challenge. Few studies have been conducted for Arabic sentiment analysis (ASA) compared to English or other Latin languages. In addition, most of the current studies on ASA have been conducted on data sets collected from Twitter and very few studies on user comments in Mobile applications (apps) reviews in Google Play Store. Therefore, this paper presents a new approach to sentiment analysis in Arabic text based on the mobile app comments dataset of Google Play Store. The proposed approach uses algorithms such as the Levenshtein distance (LD) algorithm for preprocessing the data. Thereafter, it applies various classification models to identify the mobile applications(apps) reviews in the Google Play Store of Arabic text. The results of the experiment show that the proposed approach is effective for sentiment analysis of Arabic text. The experiments were carried out by comparing the accuracy using the Naive Bayes (NB) algorithm, and we obtained an accuracy of 95.80% and compared it with the use of the LD Algorithm, where we obtained a better accuracy of 96.40% to identify the reviews of the sentence when k=9.