Twitter Ontology-Driven Sentiment Analysis

As the usage of micro-blogging services has rapidly increased in the last few years, services such as Twitter have become a rich source of opinion information, highly useful for better understanding peoples' feelings and emotions. Making sense of this huge amount of data, would provide invaluable benefits to companies, organizations and governments alike, by better understanding what the public thinks about their services and products. However, almost all existing approaches used for social networks sentiment analysis are only able to determine whether the message has a positive, negative or neutral connotation, without any information regarding the actual emotions. Besides, critical information is lost, as the determined perception is only associated with the entire tweet and not with the distinct notions present in the message. For this reason, the present paper proposes a novel semantic social media analysis approach, TweetOntoSense, which uses ontologies to model complex feeling such as happiness, affection, surprise, anger or sadness. By storing the results as structured data, the possibilities offered by the semantic web technologies can be fully exploited.