What Your Username Says About You

TL;DR

This paper investigates how much gender and language information can be inferred from usernames using unsupervised morphology induction, showing that morphological features outperform character n-gram baselines.

Contribution

It introduces a method leveraging unsupervised morphology induction to extract features from usernames for demographic inference, demonstrating improved accuracy over simple baselines.

Findings

Morphological features outperform character n-gram baselines

Gender and language can be inferred from usernames with reasonable accuracy

Unsupervised morphology induction effectively captures meaningful sub-units in usernames

Abstract

Usernames are ubiquitous on the Internet, and they are often suggestive of user demographics. This work looks at the degree to which gender and language can be inferred from a username alone by making use of unsupervised morphology induction to decompose usernames into sub-units. Experimental results on the two tasks demonstrate the effectiveness of the proposed morphological features compared to a character n-gram baseline.