Extracting important tweets of a user: A rough set approach

Microblogs such as Twitter are user driven content generation systems that allow users to share short text messages on a variety of topics such as daily conversations, news, personal updates and URLs of interest. As users continue to post, their past tweets generate a history of the shared events, opinions and reactions. Some of these tweets convey information that are of mass appeal and evoke more discussion. These tweets have a stronger social impact and may be considered to be more important than others from the authors' perspective. Authors will be greatly benefitted if an automated mechanism is developed to categorize their posted tweets and extract the most important ones. In this paper, we identify the parameters that reflect the impact of a tweet for example, the length of time during which it sustained listeners' interest (time impact). We employ topic detection techniques to determine the semantic interrelationships between tweets and retrieve the underlying theme in the thread of discussion. We propose a Rough Set based rules generation technique to sieve out important tweets along a user's timeline and demonstrate the results.