Semi-automatic detection of Sentiment Hashtags in social networks

Abstract

Communication in social networks is often carried in messages of limited size, and in some cases, like Twitter, the limit is imposed by the social network itself. Therefore, it can be hard, even for a human expert, to calculate the correct sentiment based on the message alone. Sentiment hashtags present one way to help determine tweet polarity better. In this paper we present a way to semi-automatically detect sentiment hashtags from initial tweet sentiment analysis and then recalculate tweet polarity and improve accuracy. The methodology presented helps jumpstart sentiment analysis research.