“…In recent years, machine learning-based methods for sentiment classification have been widely adopted due to their excellent performance (Li, Wang, Zhou, & Lee, 2011, 2012Xia, Wang, Hu, Li, & Zong, 2013;Ye, Zhang, & Law, 2009;Yin, Wang, & Zheng, 2012 key issues in machine learning-based methods (Zhai, Xu, Bada, & Peifa, 2011b). Up to this day, a variety of feature extraction methods have been proposed, including single words (Tan & Zhang, 2008), single-character n-grams (Raaijmakers & Kraaij, 2008), multi-word n-grams (Li & Sun, 2007), lexical-syntactic patterns and many other novel models (Xia & Zong, 2010;Zhai, Xu, Jun, & Peifa, 2009;Zhang, Liu, Lim, & O'Brien-Strain, 2010).…”