Arrive at Netflix headquarters.

Zhou Buqi saw an acquaintance, Qu Hanhan, an early employee of Youyi.com, a top student in the computer department of Beihang University, and Ma Pingshan's college friend.

Zhou Buqi didn't have much contact with him, but he left a deep impression. Firstly, his name "Qu Hanhan" looked a bit like a girl's name, and secondly, he was very courageous. At the beginning of last year, he volunteered to come to the United States for a business trip.

He is now at the senior director level of m5. He is involved in the personalized advertising recommendation system of Friends Network and is considered the number two person.

Personalized recommendations will be a bad theory ten years from now.

But at this time, it is still a very new thing.

The famous Facebook's innovation at the application level mainly has two points, one is the open platform, and the other is the recommendation of personalized advertisements.

With the application of Facebook's personalized advertising, the advertising model in Silicon Valley is increasingly moving towards personalization. The leaders among them are of course Google, Facebook and Ziweixing International.

The technical director of Ziweixing International's personalized business is Qu Hanhan, who is temporarily seconded to the United States. Ziweixing has developed in-depth in this field, so Ziweixing International's personalized advertising business has developed relatively smoothly.

In the past half year, Qu Hanhan, under the arrangement of Zhou Buqi, completed a "second secondment" to help Netflix research related personalized recommendation systems.

The results are remarkable.

Jason Kilar didn’t quite understand this, “Doesn’t Netflix have no ads? What are the personalized recommendations?”

"Film."

Netflix CEO Hargentins maintained a winner's smile on his face.

In the past few years, Netflix and Hulu have been competing fiercely and have had many wars of words. Now, the situation has become clear. Hulu’s core team has resigned en masse, and even Jason Kilar is desperate to provide Netflix with services.

Served.

It feels so good!

Of course, he also knew that the reason why all of this went so smoothly was not because of how powerful Netflix was or how excellent the team was, but because of the series of operations behind the scenes by Boss Zhou.

In the current United States, in the field of long-form video streaming, the most well-known ones are Netflix and Hulu. Boss Zhou turned his hands for clouds and rain, easily defeated Hulu, and then mobilized Hulu's core strength

Transferred to Netflix camp.

This made Hargentings feel excited and at the same time full of respect for Zhou Buqi.

This is the real boss!

A foreigner bravely breaks into Silicon Valley and can overcome all obstacles. What a heroic figure!

"Movies?" Jason Kilar was not very understanding. "Personalized movie recommendations?"

Hargentings smiled, "Yes."

Jason Kilar didn't believe it, thinking it was a difference in understanding of terminology between the two parties, and repeated, "Is it a personalized recommendation, not a typed recommendation?"

Typed recommendations are very simple, and Hulu can also do it.

For example, if a user likes to watch action movies, the system can continue to recommend other action movies to him. If the user likes to watch horror movies, the user can recommend other horror movies to him.

The core of typed recommendations is based on movie classification and relies on objective data. Personalized recommendations are different, based on user classification and relying on subjective judgments of algorithm technology.

But, can computer technology really do it?

Traditional TV programs, including film and television program recommendations on Hulu, rely on real-life data, such as TV series ratings, movie box office data, DVD market sales data, etc.

There are also movie rating websites.

These are real data and exist objectively.

The more objective, the more rational.

For example, if a movie does very well at the box office, then it must be right to recommend it to users. For example, if a TV series has very good first-run ratings, then it must be right to recommend it to users.

If it is a subjective judgment based on computer technology, what is the logic of this operating model? If the subjective recommendations of the algorithm and the typed recommendations relying on traditional data recommend the same film and television works, then this so-called artificial intelligence personalization

What is the meaning of recommendation?

After spending so much money, it seems ridiculous. In the end, shouldn't we rank the film and television works from high to low based on the movie box office, DVD sales and TV ratings?

Just like if a user likes to watch love movies, then just recommend them in order from high to low according to the classification of love, such as "Titanic", "Gone with the Wind", "Roman Holiday", "Edward Scissorhands", etc. What kind of personality is needed?

change?

Hagentins felt a sense of superiority in his heart and kept a polite smile. "It is indeed a personalized recommendation. This is a new idea provided to us by Mr. Zhou."

"this……"

It was difficult for Jason Kilar to understand, so he looked at Zhou Buqi.

Zhou Buqi nodded, "To put it more broadly, this is the gap between expected preferences and actual preferences. The expectation is to stay healthy by not smoking or drinking, but the reality is to both smoke and drink; the expectation is to start a business to change the world, but the reality is that after graduation

I took the civil service exam; the expectation is to travel around the world, but the reality is daily necessities. Everyone has a better expectation, but everyone has their own special reality, which is personalization."

Hargentins couldn't help but applaud.

Jason Kilar still didn't understand and looked confused.

Zhou Buqi smiled and said, "A movie with a high box office shows that the movie is in line with the most popular tastes. A movie with high ratings shows that the movie is in line with the public's aesthetics. But every different individual has some personal characteristics."

Preference, for example, I like suspense movies very much, even many have low ratings and low votes

I am also willing to watch suspense films about the house. For me, suspense films with low ratings and low box office are more attractive than the best romance films. The traditional type recommendation model is biased towards the taste of the public. Artificial

Intelligent personalized recommendations can analyze each person's true preference characteristics."

Just like everyone knows that "The Godfather" is the greatest work, in ideal expectations, all movie fans should watch such a work. However, on a free afternoon on a certain weekend, many people would rather choose to watch "Tiny Times"

, nor will I watch "The Godfather".

Jason Kilar suddenly realized it and gave his own understanding, "For the platform, users have two preferences. One is the preference they want to achieve, and the other is the preference actually observed. The user's choice may not be

His true self may be his misunderstanding of himself. Personalized recommendation is to give the movie that best suits his true preferences based on the observed user behavior. This has nothing to do with high or low box office, but with high ratings.

, low ratings have nothing to do with it, every user actually has bad movie attributes based on personalization.”

Zhou Buqi smiled and said: "Yes, no matter how exciting American football is, I don't watch it. No matter how bad our national football team is, as long as I have time, I will not miss a game. This is personalization, and it has nothing to do with whether it is good or not.

It has to do with one’s actual preferences.”

Jason Kilar took a deep breath. Unexpectedly, Boss Zhou and Netflix had reached such a high level in the theoretical level of streaming media, and then asked: "Has it been applied?"

It was Qu Hanhan's turn to answer, and he gave an explanation in a decent manner, "It has been tried for 4 months, the algorithm is still being adjusted, and the initial results are relatively good."

Jason Kilar pressed: "How good is that?"

Qu Hanhan said with a smile: "Many users have commented, saying that they were surprised to see a movie on Netflix that they had never heard of before but liked to watch."

"hiss!"

Jason Kilar was so shocked that he was speechless.

yes!

Can users not be surprised?

The traditional recommendation model is based on the reputation of film critics, movie box office and TV ratings, and only recommends "good movies". Netflix's personalized recommendation is different, and may include some movies that were completely ignored by people in the past.

Recommend "bad movies".

This is very important!

Traditional platforms, whether it is the DVD disc market or the TV broadcast market, generally sell "good movies", but such good movies are almost impossible to obtain on streaming media platforms.

Hulu can’t get it, and Netflix can’t get it either.

Most of them can only accept bad movies.

But bad movies often mean there is no market and no appeal. If they are forced to be widely recommended to users, they will inevitably cause dissatisfaction. This kind of personalized recommendation can solve this problem and make low-priced "bad movies"

"Film" has the opportunity to realize its value.

With the general trend of streaming media, the copyrights of good movies are bound to become more and more expensive. Whoever can promote bad movies with very low copyright prices and get recognition from users will be able to stand out and have content competitiveness!

This is due to the competitiveness of artificial intelligence technology.

"A frog in a well! A frog in a well!"

Jason Kilar blamed himself and sighed.

In the past few years, I went to Los Angeles to start up Hulu. I was far away from Silicon Valley, and I felt out of touch with the times. Maybe in terms of business, there is not much difference between Hulu and Netflix.

However, in terms of strategies, concepts, etc., there has been a huge gap like a chasm between the two sides.

Behind Netflix is ​​Ziweixing Global and Boss Zhou!

At this moment, the last doubt in Jason Kilar's heart disappeared, and the last bit of pride was gone. He made the final decision to join Ziweixing Global!

I want to assist Boss Zhou to show off his skills in the streaming media industry!

Boss Zhou is definitely a great entrepreneur, strategist, and Internet theorist. He is even better and more amazing than Bezos.

Such people deserve to be helped by oneself!

Zhou Buqi also saw something from his expression and eyes. He was calm and calm in his heart. He was very confident and had been prepared for a long time, so he asked Jason Kilar and Hagentins to communicate.

He himself called his old subordinate Qu Hanhan to talk about some personalized personal matters and show his concern.

"I heard you always go to massage parlors?"

"ah?"
Chapter completed!