Emoticon Smoothed Language Models for Twitter Sentiment Analysis

Twitter sentiment analysis (TSA) has become a hot research topic in recent years. The goal of this task is to discover the attitude or opinion tweets, which typically formulated as machine learning based text classification problem. Some methods use manually labeled data train fully supervised models, while others some noisy labels, such emoticons and hashtags, for model training. In general, we can only get limited number training models because it very labor-intensive time-consuming label tweets. As with hard them achieve satisfactory performance due noise labels although easy large amount Hence, best strategy utilize both However, how seamlessly integrate these two different kinds into same framework still challenge. paper, present novel model, called emoticon smoothed language (ESLAM), handle basic idea on data, then smoothing. Experiments real sets demonstrate that ESLAM effectively outperform those using one them.