Sentiment Analysis in Arabic Social Media Using Association Rule Mining

• Main Authors: Ahmed, AL-Saffar, Bilal, Sabri, Hai, Tao, Suryanti, Awang, Mazlina, Abdul Majid, Wafaa, ALSaiagh
• Format: Article
• Language: English
• Published: Medwell Journals 2016
• Subjects: QA75 Electronic computers. Computer science
• Online Access: http://umpir.ump.edu.my/id/eprint/37903/1/Sentiment%20Analysis%20in%20Arabic%20Social%20Media%20Using%20Association%20Rule%20Mining.pdf; http://umpir.ump.edu.my/id/eprint/37903/; https://medwelljournals.com/abstract/?doi=jeasci.2016.3239.3247
• Institution: Universiti Malaysia Pahang

The fast-paced growth in worldwide webs has resulted in the development of sentiment analysis it involves the analysis of comments or web reviews. The sentiment classification of the Arabic social media is an exciting and fascinating area of study. Hence this study brings forth a new method engaging association rules with three Feature Selection (FS) methods in the Sentiment Analysis (SA) of web reviews in the Arabic language. The feature selection methods used are (χ2), Gini Index (GI) and Information Gain (GI). This study reveals that the use of feature selection methods has enhanced the classifier results. This means that the proposed model shows a better result than the baseline result. Finally, the experimental results show that the Chi-square Feature Selection (FS) produces the best classification technique with a high accuracy of f-measure (86.811).