Vanessa Macha Undergraduate Dissertation 2014/15

Author Profiling - Age and Gender Detection

Supervised by M.Stevenson

Abstract

Individuals differ from each other drastically in terms of personality and other demographic attributes, including how age and gender are expressed in different kinds of texts. Research in this field has advanced over the years with the rise of social media as obtaining the necessary data for age and gender detection analysis has become easier.

This project aims to deliver an in-depth discussion of research and relevant methods around the task of age and gender detection in preparation for a human baseline study and an approach to automatic age and gender detection.

The best way of classifying unseen data regarding age and gender detection is sought out, implemented and then compared to the results of the humand baseline study. The comparison intends to identify the differences in how accurate humans and computers are able to guess the age and gender of strangers based on only their Twitter feed.