Friday, November 25, 2011.

On the same day that the two parties reached an agreement between NBA and its workers, Blockbuster's online business, which had been abandoned for a long time, was reopened without any advance publicity.

Although he still uses the domain name of Blockbuster to rely on Blockbuster's name, the new Blockbuster website is completely different from the past. It can even be said that the relationship with the original Blockbuster is only one name left.

New Blockbuster provides three services that do not include the original DVD rental and sales business, but instead are online streaming media playback, video sharing and online live broadcast.

The online streaming service targets Netflix, who can watch more than 20,000 movies at will on New Blockbuster's website as long as you pay 3.99 magnesium per month.

Of course, there is an old saying in Huaguo that is called "Every penny get what you pay for." This statement is all-round, and it is also true in the magnesium country.

The monthly rent of New Blockbuster is cheap, mainly because the film library is too bad. Most of the content is the copyright of various old movies purchased by Blockbuster before. These movies are mainly concentrated in movies from the 1970s and 1990s.

It only accounts for about 20%, and the movies in the new century are only 10%. As for the latest Hollywood blockbusters, sorry, none of them!

The opponent that the video sharing business targets is YouTube.

In 2006, Google spent 1.65 billion magnesium to acquire the oil pipe that had been founded for only more than a year. Many people thought it was crazy at that time.

As a result, last year alone, YouTube brought Google more than 1.5 billion magnesium in net income. It can be said that Google started a particularly cost-effective business.

In 2011, the oil pipes have grown into the world's largest video website, with more than 700 million users, and users around the world spend hundreds of millions of hours on oil pipes every day.

There are countless video websites that want to get a piece of the oil pipe, and Blockbuster is just one of them.

itch, this is an e-sports video platform that was just founded in June this year. Video game players can watch other players' game situations in real time on the platform.

Li Rui copied the live broadcast plan of Jiujianfang to the New Blockbuster website. Not only did there be live broadcasts from gamers, but also opened a channel for beauty, two young men's favorite elements: games and beauty.

If you gather, you don’t believe it, you won’t attract users.

With the launch of Blockbuster's website, there is also a collaborative filtering algorithm developed by Zhang Ming,

This algorithm can group users and recommend movies in a targeted manner. In fact, it is the prototype of personalized distribution and intelligent recommendation of hot topics in previous lives.

On the surface, Li Rui is enriching Blockbuster's recommendation system, but in fact, it is to help Zhang Ming complete the algorithm research as soon as possible.

In order to accumulate enough data, New Blockbuster not only used all previous DVD movie rental and sales data from Blockbuster, but also launched a movie rating system.

Cluster users based on the overlap of positive or negative reviews, and then recommend them through the movies that users like.

In order to make this system work, Li Rui and Zhang Ming adopted gamified design. After users collect, watch, and evaluate movies, they will receive various achievement rewards. When users browse the New Blockbuster website, the system will use it.

Various ways to attract or tempt them to rate movies. As long as there are enough users to provide more reviews, the database can provide the most accurate data support, and build a collaborative filtering process that can predict users' movie preferences, weighing many trade-offs.

Recommend movies to users after key factors.

At present, this algorithm is still relatively immature and needs to undergo continuous evolution in the future.

In order to speed up this evolution process, Li Rui launched a heavy reward.

Blockbuster Grand Prize!

This is a award for machine learning and data mining, aiming to improve the prediction capabilities of collaborative filtering algorithms. Any team that can increase the accuracy of the algorithm by 10% will receive a one million magnesium bonus from Blockbuster!

The specific quantitative index is rmse - the root mean square error of the predicted score and the real score. Zhang Ming's algorithm's current rmse score is 0.9525. Whoever can reduce this score to 0.8572 or less can get a

million!

Li Rui did not do any publicity for the launch of New Blockbuster, but he also posted the news of Blockbuster Awards on Twitter, Facebook, reddit and various other online media, which immediately attracted scientists and mathematics around the world.

The concern of home and algorithm researchers.

...

Frenhan Park, New Jersey, is a ninety-minute train ride from Manhattan. You can’t see the bustling of the big city and the rush of Wall Street. There are only a large area of ​​green fields and lush trees.

Shannon Laboratory, a simplified building, is located here. It is filled with early telephones and various electronic devices that can be called antiques.

Chris Wolinski is the executive director of the laboratory. He is sitting in the lounge on the second floor fiddling with his cell phone to relax his brain, which is occupied by various data. The reddit page that has just been refreshed suddenly pops up.

An eye-catching title.

[Do you want to win the Million-magnesium prize? Blockbuster is waiting for you!]

Million magnesium?

Blockbuster?

Both of these words attracted Wolinsky's attention and immediately clicked on that post.

After reading the content, he immediately logged onto the Blockbuster website. Sure enough, he found that the website, which was considered dead, was resurrected at some time. The clean and tidy pages, organized arrangements, and the three new business segments were all telling the world:

The giant that once ruled the entire country's home TV is back!

When Wolinsky was young, he was once a loyal user of Blockbuster. Of course, he had already abandoned the old Blockbuster and turned to Netflix's arms, but he was still a little excited to see Blockbuster come back to life.

What's more, he was very interested in the million-dollar prize offered by Blockbuster.

You know, there is a data mining team in his lab that has been predicting potential behaviors of customers for many years.

Data mining is actually an algorithm, a process of finding predictive or critical models in huge data sets.

For example, through tens of millions of searches on Google pages, can we determine which customers are more likely to buy Apple phones?

Or through data analysis on credit card consumption, customers who judge what kind of behavior are more likely to become fraudsters?

Or after analyzing data from a certain bankrupt group, what are the most likely risks to be faced by similar groups?

Wolinsky felt that the Blockbuster award was tailor-made for his data mining team. He immediately sent a group text message to more than 20 members of the team on his mobile phone, telling them the link and content of the award.

Soon, one reply after another returned like snowflakes.

"Come on, let's compete for the million-dollar prize together!" said an employee named Bell.

At the same time, many talented people have made such a voice in the country of magnesium, Europe, and many places around the world.
Chapter completed!