Analysis of Twitter Lists as a Potential Source for Discovering Latent Characteristics of Users

Abstract

We discuss our findings from a study using Twitter lists to infer the characteristics and interests of users. Gathering and structuring user interest has been challenging because it often requires expensive and/or proprietary data such as users' clickthrough logs or desktop histories. We show that by using the tweets of all the users in a Twitter list, we can discover characteristics and interests of the users in that list, even if the users as individuals do not tweet about those interests. We conducted an experiment in which we compared the user interests as found by our system using Twitter lists with those that are perceived by the human subjects in the user survey. The survey confirmed that Twitter lists reflect well the perceived characteristics and interests of the users in those lists. The user survey also confirmed that the words extracted from each set of lists are representative of all the members in the list even if the words are not used by those members.