Abstract
Sentiment analysis from Twitter is one of the interesting research fields recently. It combines natural language processing techniques with the data mining approaches for building such systems. In this paper, we introduced an efficient system for Twitter sentiment analysis. The proposed system built a machine learning model for detecting positive and negative tweets. This model used different techniques to represent the input labeled tweets in the training phase using different features sets. In the classification phase, the classifier ensemble is presented with different base classifiers for more accurate results. The proposed system can be used for measuring users’ opinion from their tweets which is very useful in many applications such as marketing, political polarity detection and reviewing products.