Back
Ioannis Pitas (AUTH)
Author/s
About the resource/s
Since the formulation of states, political authorities and political parties have sought to gauge public opinion. In recent years, opinion polls have been able to gratify this need rather efficiently. However, they are expensive to run, and their results may be biased primarily due to improper population sampling. As social networks are used on political dialogue fora, there is an opportunity to engineer automated processes of measuring the public opinion using social network texts, e.g., political tweets. We propose an innovative formula for combining tweet sentiment analysis’ results. In addition, we introduce a hybrid way to regress poll results from tweets, which enables more accurate and frequent public opinion estimation predictions.
Media
