User-annotated microtext data for modeling and analyzing users' sociolinguistic characteristics and age grading

Information from Twitter messages have become an important area for research in computational analysis of natural language. As yet, much latent user attribute analysis on Twitter is unexplored. One reason is that only few latent attributes are explicitly defined by users on Twitter. This work presents and analyzes a data set annotated by Twitter users themselves for age and other useful attributes for use in latent attribute inference applications. We report on statistical analysis of the collected latent attributes and tweet information using association mining.