Automatic Estimation of Bloggers' Gender

Daisuke Kobayashi, Naohiro Matsumura and Mitsuru Ishizuka

We propose an approach employing Support Vector Machine (SVM) to estimate bloggers’ gender from blog posts. The data we analyze consists of blog posts on Doblog (Japanese blog-hosting service) and questionnaire results by Doblog users. Experimental evaluations show that our approach achieved 90% accuracy for 83% bloggers.

Poster

Available as PDF