Opinion Mining of Twitter Events using Supervised Learning

Abstract

Micro-blogs are a powerful tool to express an opinion. Twitter is one of the fastest growing micro-blogs and has more than 900 million users. Twitter is a rich source of opinion as users share their daily experience of life and respond to specific events using tweets on twitter. In this article, an automatic opinion classifier capable of automatically classifying tweets into different opinions expressed by them is developed. Also, a manually annotated corpus for opinion mining to be used by supervised learning algorithms is designed. An opinion classifier uses semantic, lexical, domain dependent, and context features for classification. Results obtained confirm competitive performance and the robustness of the system. Classifier accuracy is more than 75.05%, which is higher than the baseline accuracy.