Social media platforms, such as Twitter, provide a forum for political communication where politicians broadcast messages and where the general public engages in the discussion of pertinent political issues. The open nature of Twitter, together with its large volume of traffic, makes it a useful resource for new forms of ‘passive’ opinion polling , i.e. automatically monitoring and detecting which key issues the general public is concerned about and inferring their voting intentions. In this paper, we present a number of case studies for the automatic analysis of UK political tweets. We investigate the automated sentiment analysis of tweets from UK Members of Parliament (MPs) towards the main political parties. We then investigate using the volume and sentiment of the tweets from other users as a proxy for their voting intention and compare the results against existing poll data. Finally we conduct automatic identification of the key topics discussed by both the MPs and users on Twitter and compare them with the main political issues identified in traditional opinion polls. We describe our data collection methods, analysis tools and evaluation framework and discuss our results and the factors affecting their accuracy.
Via Ashish Umre