Sentiment Mining Using Ensemble Classification Models

We live in the information age, where the amount of data readily available already overwhelms our capacity to analyze and absorb it without help from our machines. In particular, there is a wealth of text written in natural language available online that would become much more useful to us were we able to effectively aggregate and process it automatically. In this paper, we consider the problem of automatically classifying human sentiment from natural language written text. In this sentiment mining domain, we compare the accuracy of ensemble models, which take advantage of groups of learners to yield greater performance. We show that these ensemble machine learning models can significantly improve sentiment classification for free-form text.