Acquaintances and dating are eternal, they will always exist in various forms: on the street, on the internet, on websites, or inside mobile applications. People will always have a desire to get to know someone, and everyone will choose the most convenient way for themselves. Some continue to get acquainted in bars, cafés, and on the street; others use modern technologies and get acquainted online.
How it works
If you do not mind scammers and spammers in this segment, then in fact, on websites like Ladadate – Russian brides or in mobile applications, people are looking for a soulmate, and their interest in each other should be mutual. Anything that does not meet this dating criterion is, in fact, a waste of time for one of the parties.
The most established approach in terms of presenting information is scrolling, where one profile is displayed and the viewer must decide whether he/she likes it or not (like, dislike). If the profile is pleasant, then a notification goes out to her, and if the interest in each other is mutual, then it becomes possible to start a dialogue. There may be deviations from this dating scheme in different applications and on different sites, but the general principle of operation should be clear.
What exactly is the problem?
On average, according to statistics, there are 17 likes for every 100 dislikes. That is, a person does much more useless work than useful and receives content that is not of interest to them. This is taking into account the current parameters for filtering the dating results (search by city, age, interests, and other characteristics specified in the profile). It turns out that the primary action of a person is associated solely with the perception of the presented picture. Further assessment of the interlocutor will be carried out at the next step when mutual sympathy has been formed.
The ultimate goal is to reduce the number of useless user actions and make a selection of profiles in dating better than at the current stage.
What if everyone is similar?
The assumption and hypothesis are as follows: if one dating person likes someone, taking into account basic search filters, such as city, age range, and gender, then the results (selection) of another one who likes this same person will be more likely to match the first one and, taking into account repetition of hypotheses and increase in the number of people, the quality of the sample will improve.
An extension of this hypothesis is the minimization of incoming filtering parameters to gender and city, as well as taking into account not only outgoing likes of the desired people with similar preferences but also the responses to their profiles, which they like.
Teaching the neural network to decide for us
The assumption and hypothesis sound like this: perhaps the face in the photo plays a more significant role in choosing people than the general background, location, accessories, quality of shooting, and other factors.
The advantage of this option over the first one is the high speed of searching for preferences by pictures and the minimum number of queries and dependencies at the SQL query level.
Now the stage of test operation of both solutions is underway and the algorithms are provided to users selectively.