Platonic friend dating site
Daher says Patook built its flirt-detecting, messaging parsing algorithm by training it on sentences considered flirty (some culled from Reddit) and others considered normal.“This was done by crawling some image submissions from various sites, and running them through Optical Character Recognition,” he says.“Except for the major hubs most cities only have a few hundred users who know of the app right now so that makes it less likely for someone to find a genuine connection other than ‘they just live in my city’,” he concedes.Discussing the users it does have, Daher says the “main” demographic so far are women in their twenties or thirties.
In terms of specific steps Patook is taking to try to warm things up and get users chatting and virtually bonding, he says they’re dabbling in gamification for starting conversations.
Point being: If you’re starting from a position of there not being enough users in the first place, requiring your (very small) pool of users to earn and spend points to do things like boost their visibility seems a bit pointlessly heavy-handed (even if it’s intended to encourage greater usage overall).
Daher agrees that lack of users generally is a big barrier for Patook.
“All the above is put into a Bayes probability classifier that keeps a working probability of a user being there to flirt,” he adds.
Tech Crunch downloaded the app to test it out, and after about a week of (fairly light) usage I can safely say I have received no unwanted advances, neither in photo nor text form.