Enhancing the Performance of Sentiment Analysis Supervised Learning Using Sentiments Keywords Based Technique

摘要

Sentiment Analysis (SA) and machine learning techniques are collaborating to understand theattitude of text writer, implied in particular text. Although, SA is an important challengingitself, it is very important challenging in Arabic language. In this paper, we are enhancingsentiment analysis in Arabic language. Our approach had begun with special pre-processingsteps. Then, we had adopted sentiment keywords co-occurrence measure (SKCM), as analgorithm extracted sentiment-based feature selection method. This feature selection methodhad utilized on three sentiment corpora using SVM classifier. We compared our approach withsome traditional methods, followed by most SA works. The experimental results were verypromising for enhancing SA accuracy.