Awareness promoting visualization of biasedness of discussion progress of microblogs

Abstract

Microblogs like Twitter presents users with the others postings they choose to follow by themselves and related postings that are re-posted by the others. It raises a problem that they see only what they think favorable, although there is a wide variation of opinions. To cultivate a better information literacy, such biases should be presented to promote awareness on the biasedness not only of what they see and read but also of themselves. This study proposes a new method to detect the biased progress of discussion on a given topic on Twitter without a prepared set of keywords or dictionaries, and to visualize the pattern of the discussion progress. We employ the principal component analysis to capture the pattern of biasedness, and our contrastive experiment with human judgment shows that our method captures the real biasedness effectively.