Gender Classifier

Determine anyone's gender automatically by filling in a Twitter username, first name or an image url.

This demo has been created as part of the UN Global Pulse project on gender classification of social media accounts together with Leiden University's Centre for Innovation. The code behind this demo is used to classify the gender of more than 50 million Twitter users. The results of this project can be viewed on http://post2015.unglobalpulse.net.

A live version of the demo page can be found over here http://gender.peaceinformaticslab.org.

This code makes use of the GenderComputer of University Eindhoven, see their Github repository: https://github.com/tue-mdse/genderComputer .

Installation

Within this github there is a gender classifier which is in the folder gender_classifier and a demo page which is in the folder demo_page. See the subfolders for more information and instruction on how to install the code.

How it works

The algorithm has several input methods, which can be just the name, url or Twitter user.

By name

By just looking at the name, the algorithm verifies the name on a global scale and returns the most occurred gender based on that name.

By url

By url it identifies if there is only one face in the image. If so, it returns the gender of that image.

By Twitter user or Tweet id

It looks up the gender by just the name as in section By name.

If there is no result it looks up the gender by using the profile picture as described in section By url.

Project Team

The core team behind this project includes:

• Leiden University’s Centre for Innovation
• UN Global Pulse
• Data2X

As part of this project, valuable contributions have been made by:

• Leiden Centre of Data Science (Leiden University)
• Qualogy
• Risa-IT
• UN Volunteers
• Maral Dadvar