Title:	A SPECTRAL CLUSTERING-BASED APPROACH FOR SENTIMENT CLASSIFICATION IN MODERN STANDARD ARABIC
Author(s):	Ikram El Karfi, Sanaa El Fkihi and Rdouan Faizi
ISBN:	978-989-8533-80-7
Editors:	Ajith P. Abraham, Jörg Roth and Guo Chao Peng
Year:	2018
Edition:	Single
Keywords:	Sentiment Analysis, Modern Standard Arabic, Machine Learning, Unsupervised Approach, Spectral Clustering
Type:	Full Paper
First Page:	59
Last Page:	65
Language:	English
Cover:
Full Contents:	click to dowload
Paper Abstract:	Thanks to its various applications in a variety of domains, sentiment analysis has attracted a lot of attention in the last couple of years. Therefore, our objective in this work is to propose a spectral clustering-based approach for sentiment classification in Modern Standard Arabic. This class of graph-based algorithms has been widely successful due to its high performance in many contexts where several classical techniques fail. In fact, on the basis of the experiments we conducted on a dataset collected from Goodreads, it was found out that is this approach can automatically extract and identify the polarity of a given text and can, thus, reach higher accuracy compared to other machine learning algorithms. Results of the study have, actually, proven that our approach out performs K-means algorithms by achieving an accuracy rate of 83.13%.
	Go Back