Simple Discovery of Aliases from User Comments

Abram Handler1 and Brian Clifton2
1University of Massachusetts Amherst, 2BuzzFeed


Abstract

This work presents a new technique for discovering aliases from informal user text. We describe ongoing work towards a generative, unsupervised model for alias detection, based on a common, loose structure found in comments sections across the web. We test our method on a new corpus of user comments from BuzzFeed, a news and entertainment website.