AI is everywhere, literally everywhere.
From creating an online 3D model of your face in seconds to serve the best-suited glasses on Lenskart to eliminating production errors in large-scale manufacturing units like BMW; every industry is in love with the automation, scalability, (and convenience) that ‘AI-match’ brings to their business.
And the online dating space isn't in a vacuum either.
Tinder, undoubtedly the first-choice online dating app, is handsomely harnessing the bond between machine learning and artificial intelligence to run smoothly in three specific use-cases:
— Profile recommendation
— Harassment checking
— Smart photo feature
Here, we are going to gossip about the most controversial (and hands down the superstar!) element of the Tinder app: Match-Making or Profile Recommendation System.
Have you ever thought, out of 7.8 million-plus users, how Tinder decides to pick and show a handful of relevant user profiles in your swipe feed?
VecTec — The Secret Recommendation Algorithm
For that, Tinder uses VecTec, a machine-learning algorithm chained with artificial intelligence to generate personalized recommendations.
The algorithm evaluates users’ profiles based on multiple factors like agreeability, relationship preferences, degree of romantic passion, proximity or locality, height, religion, and level of openness to new experiences.
When any user makes a swipe, the Swiper’s characteristics are mapped on an embedded vector. As the algorithm recognizes similarity between two in-built vectors (in simple words when two users share similar traits), the AI system automatically cross-recommends the profiles more than often.
Additionally, Word2Vec algorithm learns from the communication styles of users like slang, dialects, and choice of words and infuses the learnings to group the swipes with similar preferences.
The algorithm feeds on the data it gathers from these large numbers of co-swipes and gets better with every co-swipe, improving the possibility of a match.
Though these machine learning models aren’t perfect on the emotional and context front, it’s much better to swipe across the users with the best possibility of sharing your interests, (instead of completely random users.)
So, next time you swipe right and find a match that converts into a ‘fun’ date, Thank the AI and ML-based algorithm from the bottom of your heart.
