Definition and multi-dimensional comparative analysis of ad hoc communities in Twitter

Abstract

We here present an early-stage prototype tool for defining and analyzing communities in Twitter. The tool takes a set of Twitter users and profiles them based on their tweets. This profiling is based on earlier work, where we map entities mentioned in tweets to Wikipedia entries, which in turn lets us profile a user based on the Wikipedia categories are related to his or her tweets. From here, we can define ad hoc topic-based communities (e.g., all users who discuss Wikipedia topic K). The tool is focused on contrast analysis, where we have baseline behavior or another community to compare against. Copyright © 2012, Association for the Advancement of Artificial Intelligence (www.aaai.org). All rights reserved.