Inappropriate tweets may cause severe damages on the authors' reputation or privacy. However, many users do not realize the potential damages when publishing such tweets. Published tweets have lasting effects that may not be completely eliminated by simple deletion, because other users may have read them or third-party tweet analysis platforms have cached them. In this paper, we study the problem of identifying regrettable tweets from normal individual users, with the ultimate goal of reducing the occurrences of regrettable tweets. We explore the contents of a set of tweets deleted by sample normal users to understand the regrettable tweets. With a set of features describing the identifiable reasons, we can develop classifiers to effectively distinguish such regrettable tweets from normal tweets.
Available at: http://works.bepress.com/keke_chen/48/
Presented at the 24th International Conference on the World Wide Web, New York, NY, 2015.