Contextual Analysis of User Interests in Social Media Sites - An Exploration with Micro-blogs

Since their inception, social media sites have evolved from
instant messaging and online networking to diverse, multi-
faceted entities that encompass the entire personality of an
individual. Recent advances in technology around mobile -
based access to social networking platforms and facilities to
update status information in real time (e.g. in Facebook)
have further allowed an individual's online presence to be
as ephemeral and dynamic in nature as her very thoughts
and interests. In this context, micro-blogging has been
widely adopted by users as an effective means to capture and
disseminate their thoughts and actions to a larger audience
on a daily basis. Interestingly, the daily chatters of a user
obtained from her micro blogs over a unique information
source to analyze and interpret her context in real-time -
i.e. interests, intentions, and activities. Rich contextual
information about users allow social networking players to
develop value-added applications and associated business
models to monetize the same.
In this paper, we gather data from the public timeline
of Twitter (one of the most popular micro-blogging sites)
spanning across ten worldwide cities over a period of four
weeks. We use this dataset to (a) explore how users
express interests in real-time through micro blogs, and (b)
understand how unstructured text mining techniques can
be applied to interpret the real-time context of a user based
on her tweets. Our findings provide evidence that social
media sites like Twitter constitute a promising source for
extracting user context that can be exploited by a multitude
of social networking applications.