Song Joong-ki and Song Hye-kyo Failed to Find the Right Person. Can Algorithms Help Us?
Finding a soul mate to accompany you for life is a beautiful wish of many people, but the reality is often cruel. To this end, dating websites and applications based on big data, machine learning, and AI algorithms have emerged one after another. Can they help single men and women solve this problem?
Is today "breakup day"?
In the morning, Sina Entertainment Weibo broke the news that Song Hye Kyo and Song Joong Ki announced a divorce mediation, and many fans burst into tears: their fairy-tale love was shattered.
According to a statement from Song Hye Kyo's agency, UAA, the reason for their divorce was "Incompatible personalities", but the real reason is still unclear. According to unreliable speculation, it may be that Song Zhongji is not suitable for Song Hye Kyo, but for Jia Ling...
How sweet it was before, how painful it is now
I remember when Song Joong-ki and Song Hye-kyo CP first came out, in everyone's eyes they wereThe Immortal CoupleOn the wedding day, the media reported the good news.There was also a fortune teller who said that the two of them were a perfect match based on their horoscopes.However, just a year and a half later, this was the outcome.
In fact, in addition to this couple, many celebrities have also experienced sweet and heartbreaking marriages. Princess Diana and Charles, Prince William and Cinderella Kate, Yao Chen and Ling Xiaosu, golden boy and jade girl Pan Yueming and Dong Jie, etc. are allThey entered the marriage hall with love at first, but eventually parted ways,It makes others sigh.
This is even more common among ordinary people. Vows like "stay together forever" and "never leave each other" always seem to expire easily.Finding the one you love and spending the rest of your life with him/her is a difficult problem in life.
More than 200 million single men and women in China are looking for a way out of being single
According to the statistics, the age of first marriage for single men and women in China is getting older and older. In first-tier cities,"Marriage at 30"It's not uncommon.
The reason behind this is that, on the one hand, the work pace is fast and the pressure is high, so there is not much time to spend on socializing; on the other hand, it is because everyone hopes to findPerfect partnerAnd grow old together, but it's not easy.
Now, there are alreadyMore than 200 millionSingle menfemale,Those born after 1995 have already started going out for blind dates.
In response to the needs of single men and women, various dating websites and apps have emerged in recent years. Both at home and abroad, they are using big data and AI algorithms to help single men and women match the most suitable "Mr. Right".
Is AI, the virtual matchmaker, reliable?
In general, the existing dating websites and applications are basically divided into types such as direct matching based on user big data, collecting user information through chatbots and then making recommendations, and matching through facial recognition functions.
Are these “matchmakers” made up of data and algorithms reliable?
- Find more suitable people based on neural network
Founded in California, USA Banihal Company, throughMachine Learning, Artificial Intelligence, and NeuroscienceResearch in this area is used to solve the problem of partner matching.
Before founding Banihal, founder Ishdeep Sawhney wanted to get married and settle down, but he said that every time he met someone recommended by friends or dating websites, he was not sure whether he could spend the next 50 to 60 years with her.
So, in order to find the perfect partner to accompany him for life, he specially made a long list of questions. As a result, he was always defined as "weird" by his dates.
And this list of questions gave him inspiration.He quit his job at Apple to start the matching website Banihal.
Many dating websites will match users with partners who meet hard conditions such as age, occupation, and salary, but this cannot achieve personalized matching, and the two people do not necessarily have similar interests and values.
Ishdeep Sawhney studies neuroscience to understand how people make decisions.I hope to match you psychologically and spiritually.
Banihal, which he created, asks new users more than 40 questions to answer when they register, similar to psychologist Arthur Aron's "36 questions to make strangers fall in love quickly", using these questions to let users get to know themselves and their preferences in depth. After users fill out a relatively comprehensive profile, the system will recommend 5 best matches for them.
According to the introduction,Banihal uses precise algorithms to ensure successful marriages through a software called Rae. Ishdeep Sawhney noted that Rae software feeds research tests to apps and websites to improve the matchmaking process. Rae is able to tap into subconscious signals. It has been upgraded with artificial intelligence and data to provide a better user experience.
In addition, Banihal’s AI-based software Rae continuously updates the user’s preferences, and the more the user interacts, the more the algorithm-based software Rae will update these features to present the most suitable matches.
In fact, this type of system is similar to shopping websites such as Amazon and Taobao.Recommendation System,Use machine learning-based clustering algorithms to identify groups of people that tend to be similar.
Similarity can be defined by behavior (e.g. what they buy or browse, how they pay, how often they buy, etc.), or by characteristics (e.g. their age, gender, location, etc.), or a combination of both. The recommendation engine will then predict a user's future choices based on what they did in the past, and will also take into account the behavior of other people who "like XX".
- Machine learning predicts your chances of being single
At the University of Waterloo, due to the extremely unbalanced ratio of men to women and the lack of social activities, a data scientist tried to use machine learning to help himself get out of being single.Click here for details）
He first analyzed the factors that determine whether or not a partner can be found.
What attributes do you need to stand out from the crowd of boys and be favored by girls?The young man tried to list the characteristics of boys, trying to find out which hypotheses could be supported by data.
In these cases, he assigns a value of 1 or 0 depending on whether the criteria are met. So, we are measuring the relationship between people's above attributes and being able to find objects.
In addition, he also analyzed the various reasons for being left alone.Fitness, glasses and confidenceThe three most significant variables are:
fitness:Those who regularly go to the gym or exercise are more than twice as likely to have a girlfriend (P value = 0.02)
Glasses:The probability of people who don't wear glasses having a girlfriend is 70% higher than those who wear glasses (P value = 0.08)
confidence:People with high self-confidence are more likely to have friends (P value = 0.09)
After that, the relationship between these variables was statistically analyzed, and the data showed that it looked like { has a girlfriend, looks confident, goes to the gym, doesn't wear glasses } There is a mutual relationship.
After doing these preparations, the young man trained a logistic regression generalized linear model. According to the explanatory variables listed above,Predict whether there will be a female vote.
He trained this generalized linear model with elastic net regularization using the glmnet and caret packages in R. He then optimized the hyperparameters using a standard grid search, using leave-one-out cross-validation at each iteration and optimizing the kappa coefficient. The final model had a cross-validated ROC AUC score of 0.673.
This means that the model is more reliable than your random guess in predicting your chances of finding a girlfriend.
However, it is understood that four months after publishing the article introducing his research, the young man is still single. I wonder how he is doing now...
To find a way out of being single, you must rely on yourself first. If that doesn’t work, then use AI.
It is said that the probability of meeting the right person in a vast sea of people isOne in five hundred million, but how many people can we meet in this life? There is a short film saying that we will meet 8,263,500 people in our lifetime, greet 39,778 people, get familiar with 3,619 people, and get close to 275 people. However, the right person may not be among these people.
But if you let a machine help you find someone, you can greatly improve your efficiency. You can find the most compatible person among thousands of people, and you may no longer have to search for him/her in a thousand places. The dating website eharmony said,Every 14 minutesSome people found true love there.
But, can machines really understand us better than we do ourselves? Will two people matched by machines spark love? According to online data, the success rate of dating websites is only about 1 in 10.
Although AI may not be reliable in helping you find the right person, it willHelp you avoid the wrong people, and reduce painful relationships. The AI tool will tell you not to date that guy with the nice sports car because he will just dump you in the end. In other words,Artificial intelligence can help you identify "bad men and bad women", to avoid emotional loss.
So, on the road to finding your single life, are you willing to find AI to be your matchmaker?