Abstrakt. eHarmony.com was the first online dating service to develop matching based on personality testing and analysis of married couples. eHarmony.com was the first online dating service to develop matching based on personality testing and analysis of married couples. Traditionaly it has used classic statistical methods to derive suitable compatibility models. Milions of users have taken eHarmony's personality profile and eHarmony is currently adopting adaptive techniques from machine learning to support this growth. I am going to discuss two areas of machine learning applications we are working on
i) content classification and
ii) colaborative filtering based attraction prediction.
I will describe some preliminary studies in photo ranking and learning from pairwise comparisons and discuss the possible deployment of such systems within eHarmony.