Personalized Application of Internet websites

In the Internet era, personalized applications, it is very strong, can help different users to find their favorite products, data shows that One-third of users will be based on E-commerce site recommendations to buy things, this is any advertising can not achieve results. The popularity of popular advertising on the media has been increasingly low on consumers, so someone has foreseen that personalized recommendation technology will become the ultimate form of advertising.

Many years ago, I saw a movie called "Who Knows the woman Heart", Hollywood big Mel Gibson plays the hero is a typical male. A bathroom shock accident suddenly let this big man gain the magical ability-"mind-reading", can easily insight around women's minds, hear their inner monologue. Even though he was scared to death by the skill at first, he became obsessed and captured his heart.

"Mind-reading" sounds strange, but some people are happy with it. As if overnight, a sudden presence of a "bosom friend" who knows all your preferences, can provide a full range of intimate guidance 24 hours, and patiently recommend those "you may be interested in ..." things, from the house, to the socks.

You guessed right, this is unknowingly encroach on the entire Internet "recommendation algorithm", not only directly to your bottom of the small secrets, but also become a site to win the user's core secrets.

When "recommended" makes people stop

Netflix uses software algorithms to recommend movies, and watercress is good at recommending "casual good music", and Goodreads is keen on recommending books ... Personalized "recommendation algorithm" has been fully applied to a long list of internet sites, from video recommendations, music recommendations, shopping recommendations, such as friends recommended.

Many netizens are infatuated with the use of "recommended algorithm" after the pleasure. "Since I used the music services of last.fm and watercress, I began to rely on the ' music recommendations ' that they offer, no longer stopping at the street CD booth, or even MP3 downloading. "Miss Zhang, who works for a PR company, became a big fan of watercress two years ago," she said. "No use before still don't understand why there are so many people listen to watercress, after use, I have to admire it recommended music is very consistent with my taste, the more music collection, it recommended the more accurate, like shoes more across the fit." However, if you collect too much music, the accuracy will be reduced, perhaps at this time, even you do not know what kind of music you really like, not to mention is a software. ”

In fact, the "working principle" of the recommendation algorithm is not too complicated. Take the same recommendation music Last.fm website for example, if you like Faye Wong, and you like Faye Wong's friends listening to Lam, Last.fm will put Lam on your playlist.

The head of the Last.fm website praises the recommendation algorithm, "We built a huge community around music, and this community helped us refine the recommendation." The recommended music is extracted from the true listening habits of more than 20 million of people. So the more times you play music, the more users on the last.fm, and the more accurate the recommended results will be. You can find that music and music often have an unexpected connection, or even inadvertently reveal your recent mood. Heard that someone lovelorn, the heart is unwilling, on the Last.fm to see the former lovers listening to what kind of song, guess their mood at this time. ”

To personalized "algorithm recommended", that is, "watercress guess you will like," Douban first practice of the three areas of life are books, movies, music. The reason is also simple, these three areas are easiest to recommend accurate. Douban founder Yang Bo said, "the most effective way to make choices for most people is from relatives and colleagues." Random one or two of recommendations, not only convey their own true feelings, but also contains the taste of your judgment and the subsequent selection of the line. They do not recommend a single child to a bachelor, nor do they bring back "naked" agents to their mothers. No matter how tall and thin, Snow White, watercress helps you find like-minded people through your favorite things, and then find more good things through them. ”

Interestingly, "recommended algorithms" also derive a number of additional benefits. The best thing to play on the Last.fm is to watch what music people are listening to. The data is very interesting and can even predict exactly what bands will be popular. Now, through the "recommended algorithm" to make predictions has made a lot of companies moved heart. Google has created a new product to try to predict the winners of the Oscar awards through search engines. It is understood that the Oscar for the past few years the best film "Bomb", "Slumdog Millionaire", "Old No" and so on, have been in the search engine before the award for at least four weeks of upward trend. However, it is clear that this prediction also needs to strengthen the "aim", search for the hot "social network" in the end is still in reality lost to the King's Speech.

It can be foreseen that with the progress of technology, more humane, more accurate "recommendation algorithm", even through the body feeling, iris, blood pressure and other data changes, mining to the user's real inner needs.

When "recommended" encounters "business acumen"

However, for the "recommended algorithm", the immediate question remains lingering-is this really an ideal "bosom friend"?

I do not know if you notice, every time in Taobao, from you search you want to buy things start, to complete the transaction to each other evaluation, the site will be in a small corner rolling recommend some "you may be interested in things." This is the "recommended algorithm" that is quietly hidden. For example, you have bought several works of Murakami, the algorithm will automatically recommend to you the author, as well as a number of Japanese writers and other works.

More and more people find it a lucrative business to speculate on people's tastes. The fact that Netflix, the film rental network, is putting 1 million of dollars into the development team to develop a film recommendation system that is better than the older version is the best proof. Now, there are many experts to the "referral algorithm" to the social network after the web2.0 era, "The biggest Dark horse" throne.

Anderson, editor of Wired magazine, put forward the three principles of the long tail theory, the first is to make everything available, and the second is to sell it cheaply; the third is to help me find it. And the 3rd is a personalized "recommendation algorithm" expertise to help users make choices in a large number of products.

Today, the three fastest-growing global e-commerce retail giants-Amazon, Staples and Netflix-have fully applied personalized referral systems. According to the market analysis company Forrester Statistics show that one-third of users will be based on E-commerce site recommendations to buy things, this is any advertising can not achieve results. The popularity of popular advertising on the media has been increasingly low on consumers, so someone has foreseen that personalized recommendation technology will become the ultimate form of advertising.

Domestic Douban also in the "recommended algorithm" of commercial applications, this year from the life of the station, second-hand trade in the community, "watercress guess you will like to buy", until some mobile phone applications have been the first to test water. According to Yang Bo, "We hope that when others help you play the game gossip, you can also help your real life." ”

"It used to be manually recommended, but it's more convenient and efficient to develop automated, intelligent systems." "Percentage point CEO Berlinson that personalized recommendation technology will be equivalent to the search engine of the Internet basic services, personalized referral Service Precision marketing platform will become the electronic commerce industry standard equipment function."

No wonder Google, which has repeatedly failed in social-networking products, has stubbornly rolled out "+1". Like Facebook's "like", if you see a favorite link in Google search results by pressing "+1", your friend will see your recommendation when they do a similar search. Now, the "+1" button will only appear on the Google search page, but Google is planning to make it appear on major sites.

When "recommended" about You and me living

Perhaps the automatic recommendation of this kind of computer can be a novelty, but when it's a mass of news like "You might be interested in," the books you might be interested in, the movies you might be interested in, the restaurants you might be interested in, the "you might be interested in ..." There are already a lot of people have been infatuated with the numerous sites on the flood of recommendations, it seems more like opening a "Pandora's box."

Netizen MARSC recently was recommended algorithm "Ray" a note. Originally, he in Jingdong Mall on the cheerfully ordered double buffer running shoes, the results page immediately listed a "Best purchase Portfolio", unexpectedly catch a smoke-free frying pan. "I faint, buy shoes with a pot!" I don't know how Jingdong was counted. ”

Some friends around also began to complain to reporters, listen to the site system recommended music, although occasionally can be pleasantly surprised, but always a tune of the loop playback is really hard not to let people produce aesthetic fatigue. "Most of the time, the site recommended music is very ' good ', but it's like I hired a DJ who only knows how to obey and flatter." ”

The book recommendation of Watercress also encountered a similar embarrassment. NET friend at every point open 1 books, "Also like ..." The list will always involve another 10 of a variety of books, 10 and 10 of the endless cycle, the end can only let people directly ignore the watercress system recommended, otherwise just look at these will consume a lot of time.

When large sums of money pile up the accuracy of the "referral algorithm", the systems that pretend to be able to discern your mind cannot guarantee the diversity and novelty of the recommendations. "The recommended algorithm will limit the areas of interest that we are interested in, prevent us from discovering new highlights", and some netizens are outspoken in the forum. When the recommendation is out of control, it makes it impossible to filter out what you might be more interested in.

Realistically speaking, improve efficiency, growth insight is not always the "recommended algorithm" the ultimate goal, the development of its Web site is the user stay more time, or spend more money, for this may even hesitate to continue to dig more personal privacy.

A little story seems to be the best proof that math genius Jeffhammerbacher,2006 graduated from Harvard a year later and joined Facebook as a cornerstone of Facebook's business--to "recommend algorithms" to ensure accurate advertising. But only two years later, Hammerbacher began to suspect life, and he resigned from Facebook in 2008. "My head is here thinking about how to get people to click on ads massively, really bad." The genius resigned after the infinite feeling.

Yes, this is the naked "recommendation algorithm".