The perfect match

A + B = heart. The confessions are carved into young birch trees and old oak trees, sometimes they glisten at summit crosses, high up on the mountain. No one can say how long the love lasts. How it comes about but already. Portals such as Lovescout, Parship or Tinder have come to grips with the mysterious mystery of their love formulas. And have opened up a business of emotions with it – and with a lot of money. Last year, the dating exchanges have implemented 210 million euros. The online dating exchanges in Germany count around 8.6 million active users. Everyone wants the perfect match.

But is there the formula of love at all? “Every single eleven seconds a single falls in love with Parship,” promises the Single-Börse from Hamburg. It sounds like you can activate the butterflies in your stomach by pressing a button on the computer. That’s the way it works: the computer as the gateway to the vast networked world provides the basis for this. You usually pass by strangers. That you suddenly get to know each other is no coincidence – but the finest data extraction.

Also interesting: recommendations are meaningless in finding a partner


People of the same kind stick together? Yes, but no

To find the right matchmaking criteria out of abundance, each portal has its own magic formula. At Parship, one draws 32 personality traits. From this an individual profile is created and automatically compared with other profiles. It is then adjusted according to wishes for proximity to the partner, how to deal with frustration or how to communicate. There are 136 algorithms at work: the more matches they find, the better.

Logical: The more you know about a person, the better you know him and can match him with another. “The severity of Partner A is compared in a characteristic of the individual expression in this characteristic at Partner B and scored with a corresponding score,” says Christiane Lénard, Head of Scientific Development and Matchmaking at Parship.

Sometimes there are characteristics that should necessarily coincide, so that it works with love at first sight. For other characteristics, however, it is beneficial for a harmonious partnership to complement each other. Finally, the points are summed up, weighted and summarized to the overall matching result. The folk wisdom “Equal and equal joins like” therefore applies just as little as “opposites attract”.

The love formula is written in the programming language Java. Machine-learning or even artificial intelligence is not used. For Lénard have recommendations on how they use online retailers à la “Other customers also bought XYZ” in the search for the perfect partner makes no sense. A dealer knows how many of the ordered shoes will be returned – unfortunately not the partner exchange.

More choice, more success

According to a representative poll by Parship and Innofact 2017, 53 percent of German singles with their desire for a relationship are no longer just looking for friends or sports clubs, but are banking on the much greater possibilities of the network. And they are good at it. So they have the largest selection and the best chance of success. Three quarters of all Parship couples are very satisfied with their current relationship. The same applies only to half of the German couples who have not met on the Internet.

Tinder: Less is more

Tinder is already using the technologies to find Mr. and Mrs. Right. The dating app analyzes which pictures are best received by the other users by changing the set photos and adjusting them via machine learning. This optimizes the chances of being liked. Machine-Learning uses the basic logic familiar from A / B testing, but in return can test several variants and adjust them in real time. Also, the love nose of the AI ​​is in use: The feature “Super Likeable” a user four different candidates are displayed, to which he can send a Superlike. The other one will be notified immediately. An AI has previously matched their preferences.

Who on the principle “a lot helps a lot” sets, however, has bad cards in the love matcher. Tinder has introduced the Elo score. The algorithm ensures that the principle “a lot helps a lot” does not prevail. Men, in particular, apparently give the women many expressions of affection in the form of a green heart. And hope that many random hits come out. The crooked tour is seen through the Tinder algorithm: He sorts out the Dauerliker. On the other hand, those who forgive little, but receive more, signal attractiveness. Ergo, he gets a higher credibility: he can better judge who is attractive – and who is not.

Other data is also involved. Very simple like age, gender and geolocation, but also interests and preferences. Tinder draws on a large data pool: Who logs on to Tinder, makes that usually on Facebook. There, the data are already stored. In addition, there are information about friends, as well as photos and likes for topics that are of interest. This information tells Tinder about the behavior of the user. This fences the selection of candidates who come into question for the Munich area alone, once drastically.

HR: Making the right assumptions

Finding the perfect match is not just a topic for lovers. Personnel consultants and human resources managers are also racking their brains over which employee best suits the company. The application folder with cover letter and CV, coupled with the question of where you see yourself in five years, is no longer enough. “Companies that have a sufficient amount of structured data are starting to use algorithms in recruiting,” says Joachim Diercks, managing director at the human resources agency Cyquest. The algorithms help to discover connections: This supports the recruiting process in some parts. The algorithms can test about 200 CVs in one second. This saves money and time.

It is also possible to test whether an applicant matches the company. The Talanx insurance group uses the “Precire” language software to select the best candidates for the highest positions in the Management Board and Management. A 15-minute telephone conversation is sufficient. The program provides the employer with a psychological profile of the human being by asking about 42 questions. They revolve around personal and professional matters, but do not play a role – they are meant to show how an applicant reacts. Using this method, Talanx has reduced the cost of recruiting up to 70 percent.

For Malte Balmer, this approach is not an alternative to personal selection by a human resources expert. Balmer is Talent Sourcing Manager at Otto and regularly attends trade fairs and conferences where recruiting tools are presented. He has not yet found something. “For us, the overall package and the question of whether the employee fits in with Otto are more important than grades.” This question of culture can not be filtered out of the body of applicants by an algorithm even in advance, “he says. Only one person can judge another person correctly.

M & A: Algorithms reduce mismatch

What about more abstract matchings? Merger & Acquisitions is such an area where algorithms are being used more and more often. Whether a company or a start-up fits in with a corporation is difficult to say, which is why 70 percent of all takeovers fail. A high failure rate. In order to lower them, robots are now reaching under the arms of the companies. IBM uses the M & A Pro tool, a machine learning system that uses an algorithm to evaluate historical data. Incidentally, the tool has emerged from an acquisition: IBM took over the statistics software company SPSS in 2009 for 1.2 billion dollars (about one billion euros). That has paid off according to CFO Martin Schroeter. As he told the Financial Times, M & A Pro provides more accurate insights into a potential takeover candidate, speeding up the M & A process before a competitor targets it.

The preliminary work of the robot is not in vain. Not even with the love search. Lénard refers to a study by the Federal Statistical Office, according to which the divorce rate in 2017 fell by 5.5 percent compared to the previous year. How much of it goes to the account of the dating portals, however, is in the stars. Three quarters of all Parship couples are very satisfied with their current relationship. The same applies to only half of German couples who have not met on the Internet.

Pure random? Nobody falls in love online on a dating site, but always through the actual encounter, says Carola Antelme, German market director at Meetic with his German portal Lovescout24. According to her, the online and offline worlds will mix even more closely in the future. Through the Internet of Things, suitable partners are also made aware of one another in sport and in the supermarket – and can get in touch via the technology.

This is becoming increasingly automated, says Antelme: “Apps will decide on the basis of our data and powerful algorithms independently, with which partner we flirt and they will automatically send the first flirt messages.” Algorithms and automation or not: For the Meetic manager, the portals are more of a medium of initiation. They help to make the right choice from a huge pool of singles. “You have to fall in love with yourself then.”

Lead 3 Ki Ethics 1200X1200 Vr

Human or machine

Companies experiment with artificial intelligence. Whether in customer service or medicine, AI can be used almost anywhere. Should there be limits in the application areas? In contrast to humans, the intelligent machine lacks one thing – the conscience. The discussion on ethical issues can not be avoided in LEAD Bookazine 3/2018.

Order now

Leave a Reply

Your email address will not be published. Required fields are marked *