"VIBES: Visualizing Changing Emotional States in Blogs." by April Wensel

(download April's senior project writeup)

By creating new modes of interaction between humans, the Internet provides new opportunities for the exchange of emotional experiences. Online journals (blogs), for example, provide notonly an outlet for emotional self-expression, but also a space for social interaction and commiseration. However, the massive extent of the blogosphere can overwhelm users with data and restrict their ability to make meaningful connections to fellow bloggers. Recent research into this problem has focused on methods of extracting global trends in opinion across the entire blogosphere, but there has been considerably less work on automatically summarizing content from bloggers individually. Therefore, a need exists for a system that can organize and summarize personal blog data into a manageable, accessible format.

We have developed such a system, VIBES, that extracts the most important topics from a blog, measures the emotions associated with those topics, and generates a suite of visualizations that summarize those emotions and how they evolve over time. In preliminary user tests, a majority of participants said they would find it useful to see at least one of the visualizations for the blogs they read or write. Most also agreed that each visualization either revealed the author's current emotional state or emotional development over time. VIBES has potential applications both in sharing emotional profiles among online friends as well as facilitating self-reflection through private user status displays. It also offers a fresh perspective for studying emotions and modeling how they change over time, which has a number of applications in affective computing, including the creation of emotionally responsive interfaces.