Twitter Sentiment Analysis - Exploring the Effects of Linguistic Negation

Abstract

Twitter sentiment analysis, the process of automatically extracting sentiment conveyed by Twitter data, is a field that has seen a dramatic increase in research in recent times. This Master's Thesis presents a study of the effects of linguistic negation on Twitter sentiment analysis.

Current state-of-the-art solutions in Twitter sentiment analysis and negation scope detection have been explored. Furthermore, a corpus of English Twitter data (tweets) annotated for linguistic negation has been created, and an improved system for negation scope detection in Twitter sentiment analysis has been developed and evaluated.

Our research represents the first work that explores sophisticated negation scope detection methods on tweets. The system produces better results than what has been reported in other domains. It has been incorporated into a state-of-the-art Twitter sentiment analysis classifier and the effects have been compared to other solutions commonly used within this field.

The study shows that the inclusion of the developed negation scope detection method in a Twitter sentiment analysis system improves the performance on tweets containing negation. However, the sparse distribution of linguistic negation in tweets results in a marginal performance gain on general data.