Application of Rough Set Theory to Sentiment Analysis of Microblog Data

Abstract

Microblogging has become a popular medium for broadcasting short text messages of 140 characters or less through social networks. There are millions of such posts shared each day, publically expressing sentiment over a variety of topics using popular Internet sites such as Twitter.com, Plurk.com, and identi.ca. Most of the existing works have focused on polarity sentiment analysis of these data by applying machine learning algorithms. In this chapter, we show that the rough set theory introduced by Pawlak provides an effective tool for deriving new perspectives of sentiment analysis from microblogging messages. More specifically, we introduce the use of rough set theory to formulate sentimental approximation spaces based on key words for assessing sentiment of microblogging messages. The sentimental approximation space provides contextual sentiment from the entire collection of messages, and it enables the evaluation of sentiment of different subjects, not in isolation, but in context. Sentiment, itself, is subjective. The degree of emotion that a word invokes in one person will be different than in another. It is for this reason that sentimental approximation space offers potentially more insightful information about a subject than simple polarity answers of positive or negative. © Springer-Verlag Berlin Heidelberg 2013.