Almost every Millennial/teen of this decade has experienced Tinder or similar dating apps while just trying it out for fun or while looking for companionship. The usual scenario in these apps is that the female users are flooded with irrelevant matches and therefore left swipe almost every guy. On the other hand, guys like me start by selectively right swiping people and when you realize you still haven’t gotten matched to anyone after some finite right swipes, you just right swipe everyone out of plain desperation and even find innovative ways of doing that.
The reason for these irrelevant matches is that these dating apps use rule based matching to match people based on rules like interests, age, distance etc. So the matches shown are based on how well the users are similar to the matches, based on these rules.
And one more thing with these apps is that all users, including me, are guilty of swiping profiles just based on looks. In the end, it all boils down to one thing when it comes to swiping people which is how good looking the person is according to you. Guess what. Current dating apps don’t take that into account either when showing matches.
Dating apps do not take people’s preferences into account when showing matches.
Artificial Intelligence aims to eliminate this random matching by showing the ‘right’ people for every user.
What if the dating app could understand what you want. Let me rephrase that. What if it could learn what you want based on how you behave with your current matches and predict the matches you would want in the future.
For instance, let’s take the case when the user is swiping through profiles. You may have specified that you are interested in only people who are interested in Master’s Degree. But it’s often the case that you make an exception for somebody because they had other characteristics which compensated for a master’s degree. This is where Machine learning works it’s magic. Based on this behavior, the recommender system also matches you with people with similar characteristics( No Master’s degree but other highly compensating characteristics). Gradually, the more you use the dating app, the more the machine knows about you and thus, better matches for you.
All is good and well when you get matches who you find attractive. But what is the point if it’s only one sided ie. s/he doesn’t find you attractive. You wouldn’t get a response if your match doesn’t find you attractive.
This is where the Attractiveness score becomes useful.
What is this attractiveness score?
Attractiveness score is a cross feature which is calculated based on a set of attributes such as confidence levels, matching likes, response rate etc. The features used to calculate this score will depend on the kind of data that can be extracted from a dating app. This attractiveness score will help find the kind of people who would find you attractive and thereby increasing response rate.
I saved the best for the last. Wouldn’t it be awesome if you could see profiles which You find attractive based on looks? Well, you can.
Using Neural Networks, the dating app can figure out the facial preference profile for each user based on the profiles the user has liked in the past. Using this facial preference profile, along with the behavioral data, the dating app can now recommend the best matches to the user.
Apart from these, there are other relevant features which can be used to make the recommendation engine more efficient, but it would vary with different dating apps.
We at, Marax AI, are currently helping Datings Apps/Websites integrate Artificial Intelligence so that they show relevant matches and keep their customers happy.
Powered by Marax’s AI Engine, which uses a combination of advanced machine learning, deep learning, natural language processing and other proprietary algorithms, it can help dating apps recommend personalized and accurate matches for every user.
Also, if you’re interested, you can check out our demo on how we integrate Artificial intelligence into a Dating App.