Dating App Provides User Data For Experimental Matching System Based Exclusively on Looks

Artificial intelligence researchers from the UK have worked with a dating app to create a match recommendation system based entirely on physical attributes and attraction.

The resulting product is called the Temporal Image-Based Reciprocal Recommender (TIRR). It is believed to be more accurate than less superficial ones that take into account other factors, such as age, profession and interests.

The unnamed dating app, which is described as “popular”, provided the team with a subset of data from 200,000 users, split evenly between men and women. The information was also only representative of heterosexual couples.

It was then used to train the TIRR and calculate the probability of a match between two profiles, only taking into account the available profile photos.

The results gave slightly better accuracy than a project from 2019, that followed a similar process but was based around the text aspects of dating profiles.

Martin Anderson, an expert machine learning, AI and big data journalist, noted that the results of both experiments followed a similar pattern, meaning that there is a correlation between the importance of images and bios on dating profiles.

He speculated on Unite.ai that there are two possibilities for the similarities: “[Either] users’ perception of visual attractiveness is influenced by the text content of profiles; or that text content receives greater attention and approbation than might have occurred if the associated picture was not perceived as attractive.”

Anti-superficial dating is an emerging industry trend over the last few years, which is trying to get users forming connections before seeing what each other looks like. However, the results of this research shows that physical attraction has a role in the online dating journey.

Read more here.