Emotion–Location Mapping and Analysis Using Twitter

The ever-increasing amount of text generated by Twitter users contains a wealth of information about the users' state of mind. Over the years, researchers have tapped upon this resource and proposed a number of lexicons and techniques for analysing the polarity of sentiments expressed by tweets. However, we need to delve deeper to extract the emotions conveyed by them — a research direction that had not received adequate attention so far. Through this work we develop a novel Emotion Analysis lexicon that was compiled by integrating information from the domain of psychology, the lexical ontology WordNet, and a set of emoticons and slangs commonly used in web jargon. We use this lexicon to find the predominant emotions carried by tweets originating from three different cities and analysed how they evolve with time. We propose a mathematical characterisation of the evolution of emotion by introducing the concepts of Emotion Intensity and Emotion Trend and visualise them in the context of causative events. On analysis, we observed Joy to be the predominant emotion in all the three cities. However, we could discern significant differences in the responses, patterns of emotion swings and correlations between different emotion categories that reflected the uniqueness of each city.