Online dating grows ever more popular in our digital world.
Dating sites are far more effective if they are capable of matching up people who are actually likely to talk to each other. But the goal of finding good matches is a difficult one.
Recently, a research team led by Professor Kang Zhao at the University of Iowa has developed a better algorithm for dating sites to link up singles.
Matching heterosexual couples on a dating site is in many ways similar to matching users to movies on Netflix, or matching buyers to products on Amazon. We have two sets — men and women, users and movies, buyers and products — and we want to find a way to appropriately match up members of the first set to members of the second set.
There is, of course, a glaring difference between dating and the other matchings — the "targets" being chosen are human beings, and they can choose whether or not to reply. If I want to watch "House of Cards" on Netflix, Kevin Spacey cannot say no to me. If I message an attractive woman on a dating website, it is up to her whether or not to write a reply message.
Sites like Netflix and Amazon use a process called collaborative filtering to make movie or product recommendations. The algorithm first compares me to other users, seeing how much overlap there is between the movies I watched and rated highly, and the movies that the other users watched and rated highly. This gives me a similarity score with other users — someone who, like me, has recently watched a lot of Star Trek on Netflix will have a high similarity score to me, whereas someone who exclusively watches romantic comedies from the 90s will have a very low similarity score to me.
Next, to make recommendations to me, for each movie that I have not seen, the algorithm calculates a score based on how that movie was rated by people with high similarity scores to me. Netflix recommends movies that were highly rated by people who like similar movies to me.
In the online dating context, an algorithm can get a good idea of my taste in partners by doing a similar comparison of me to other male users. Another male user of the site will have a similar taste in women to me if we are messaging the same women.
However, while this gives the algorithm a good idea of who I like, it leaves out the important factor of who likes me — my attractiveness to the female users of the site, measured by who is sending me messages.
Zhao's crucial innovation is to combine information about both tastes and attractiveness. The algorithm keeps track of both who I am messaging, and who is messaging me. If a male user has similar taste (he is messaging the same women as I am) and attractiveness (he is messaged by the same women as I am) to me, we are scored as being very similar; if we are similar in one trait — if we have similar tastes but attract (or fail to attract) different groups of women, or vice versa — we have a moderate similarity ranking, and if we are different on both measures, we are counted as very dissimilar.
Similarly, when finding women to recommend to me, the algorithm factors in both sides of the messaging coin. Women who had a back-and-forth messaging relationship with men similar to me are ranked very highly, women who had a one-sided messaging relationship with men similar to me are ranked in the middle, and women who have had no contact on either side with similar men are left out.
Zhao and his peers tested their hybrid algorithm, incorporating both taste and attractiveness information, on an unnamed popular dating site, and found that it outperformed a number of other recommender models. The algorithm did a very solid job in recommending potential matches that, if messaged, would message users back.
While online dating, like all dating, is still a very uncertain path to finding love, innovations like Zhao's can help dating sites become ever better at matching people up with each other.
More From Business Insider