This Maths Geek Gamed OkCupid And Found His Future Wife
Chris McKinlay used data and algorithms to sift through thousands of profiles and find someone special.
This is Chris McKinlay, a 35-year-old maths Phd, who was looking for love using online dating sites, without much success.
OkCupid works by asking people a series of questions about their personality then matching similar people with similar answers using an algorithm.
So while working on his dissertation, McKinlay set up 12 fake OkCupid accounts and a built a computer program to harvest data from female members who fitted his target profile.
The fake profiles, or "bots", automatically visited the profiles of women aged between 25 and 45 in the LA area and gathered their personal data and answers to OkCupid's survey questions. Within six months he had six million answers from 20,000 women.
McKinlay even trained his bots to act human, mirroring the typing speed and click rate of a human user, after OkCupid started to shut down his fake accounts.
He had gone from having a 90% "match" with a few hundred women, according to OkCupid, to having a 90% or higher match with 30,000 women.
The breakthrough came via the use of an algorithm, K-Modes, originally developed to analyse soybean crops.
McKinlay used this to sort the 20,000 women into seven clusters of different personality types - and found there was two, one full of mid-twenties arty types and one full of slightly older professional creative women, that were just right.
Now he could a) create TWO OkCupid profiles targeted at each group, using text-mining b) answer the 500 most popular survey questions among his two favourite groups.