Micro-blogging Content Analysis via Emotionally-Driven Clustering

An affective analysis methodology is proposed, which is a lexicon-based technique, for capturing the wisdom of crowds, as well as the social pulse and the trends, through the more accurate assessment of human emotion states.

Abstract

Microblogging has become commonplace and created new methods of communication, contributing significantly to information sharing. This holds since microblogging focuses on sharing content while building social relations among people who share the same interests and/or activities. In this context, people's perception and emotions towards a specific subject is a valuable piece of information and sentiment and affective analysis play an important role. In this paper, an affective analysis methodology is proposed, which is a lexicon-based technique, for capturing the wisdom of crowds, as well as the social pulse and the trends, through the more accurate assessment of human emotion states. The methodology adopted involves the monitoring of the emotions' intensity, i.e. how strong or weak the emotional states of the published information are. The results suggest that the proposed approach manages to efficiently capture people's emotions as these were recorded in datasets derived from Twitter.