Extracting interest tags from Twitter user profiles

A large percentage of the population uses social media to communicate with family and friends. Much personal information, such as the user's gender, age and personal interests are disclosed online. This personal information can be of great value to survey analysts to better study people's behavior.

While personal information can be mined from a variety of sources, one important source that has been largely overlooked is the profiles that users write themselves on social media platforms. On many social media platforms, such as Twitter, users can describe themselves in a sentence or two. Since these profiles are written by the users themselves, they will reflect the user's background, interests and beliefs.

For example, phrases like "gamer" and "love cats" clearly describe the user's interests; phrases like "writer" reveal the user's occupation; "26 years old" shows the user's age; or the catchphrase "love life" reveals the user's attitude toward life.

For example, someone mentions in her profile that she "likes interior decoration and high heels." We will find that this special interest can also be expanded to broader topics , such as "fashion" and "style".

A sample-based study found that about 28% of Singaporean Twitter users and more than 50% of U.S. Twitter users revealed personal interests in their profiles, suggesting that mining Twitter profiles is of high value.

Intuitively, phrases like "gamer" and "writer" serve as meaningful and informative tags for users.