Crossover: 2014

Chapter 100 Google's Interest
(ps: ... Hezhang, please subscribe)

……

Because of the time difference, although it is already night in China, it is still daytime overseas.

That is to say, although Lin Hui has rested, there are still many people overseas who are busy working.

In an office at MIT's Natural Language Processing Research Center, Eve Carly is still at work.

In fact, since the advent of the generative text summarization algorithm tinkered with by LIN HUI.

Eve Carly didn't know how long she hadn't closed her eyes.

The reason why I can't sleep is not because of jealousy, but because of excitement.

It seems that it is not accurate enough to describe it as excitement. To be precise, it is an unusual excitement.

Of course, excitement is not everything, there is also some fear in addition to excitement.

But deep within fear is anticipation.

Although the mood is very fan-shaped, Eve Carly is sure that each of her moods is not without reason.

As a scientific researcher, it is naturally exciting to be exposed to an unprecedented way to solve problems.

And the more you learn about generative summarization algorithms, the more this excitement grows.

With a deeper understanding, Eve also feels more and more powerful about the person who proposed the algorithm.

The gap in strength made Eve Carly a little overwhelmed, and she unconsciously felt a little afraid.

Eve Carly seems to understand a sentence:

——The closer you get to the opponent's level, the more you understand the opponent's strength, and you will become more and more afraid.

As for expectations, it is easier to understand. Human beings are inherently curious about unknown things.

While each feeling is legitimate, it's always weird to mix several moods together.

Especially for Eve Carly, a person who basically never has any mood swings.

This feeling is even more unfamiliar.

Uh, although I don't know how to describe that strange feeling.

But deep down, Eve Carly admires LINHUI, a genius from the other side of the ocean.

What is a genius?Although Eve Carly received a Ph.D. from the School of Computing at MIT, one of the world's top computer research institutions, at the age of 25.

But Eve Carly never dared to claim to be a genius. Although she had a smooth journey, only she knew the hardships she had paid.

In her opinion, the key point of genius is not "talent", but "talent"

The words "genius is 99% sweat and 1% inspiration" are completely deceptive nonsense.

In Eve Carly's eyes, people who have gone through untold hardships and worked hard to succeed may be counted as talents.

But it's definitely not genius.Where does genius use almost strenuous effort?

Perhaps a genius also needs a little hard work, but absolutely no such kind of hard work is required.

It's as if everyone seems to know how to get out of the room and find the door, but they are helpless.

And the genius is the one who walks to the door under the blank eyes of everyone and gently pushes the door open.

Words such as "go through all kinds of hardships" and "come through hardships" can only describe ordinary people.

"Walking in the courtyard" and "lifting weights like light" are the descriptions that belong to geniuses.

And LIN HUI is a genius in the absolute sense.

When everyone faces the bottleneck of the extractive summarization algorithm and cannot find a way out of the text summarization room.

LINHUI appeared just right, and pushed open a brand new door called "generative text summarization" under the confusion of everyone.

In Eve Carly's heart, LINHUI is an idol to be admired.

……

Worship is worship, technology has no borders, but technicians have borders.

Eve Carly's team received the task of following up the technology proposed by LINHUI as soon as possible.

The task received was not issued by a higher-level scientific research management institution.

Rather it was proposed by Google/Google.

The natural language processing project of Google/Google and Eve are in-depth strategic partners.

Every year Google/Google sponsors the team over ten million dollars.

To put it bluntly, Google/Google is the father of the research team where Eve Carly works.

Google/Google's task is simple (at least the person who assigned the task at Google/Google thinks so):
——Assess the feasibility of the algorithm proposed by LIN HUI, and consider whether it can be reproduced in a short time according to the actual situation.

As for why Google/Google is interested in the algorithm proposed by Lin Hui?
This has a lot to do with the history of Google/Google.

The reason why Google/Google is today is largely due to the PageRank algorithm.

In the early days of the Internet, with the gradual increase of web pages on the Internet, how to retrieve the pages we want from the massive web pages has become very important.

At that time, the famous Yahoo/Hoo and other Internet companies tried to solve this problem, but they failed to have a good solution.

It was not until around 1998 that two doctoral students at Stanford University, Larry Page and Sergey Brin, jointly invented the famous PageRank algorithm, which perfectly solved the problem of web page ranking.

It is precisely because of this algorithm that Google/Google was born.

PageRank is a technology that calculates the importance of web pages through hyperlinks between web pages.

Named after the surname of Google/Google founder Larry Page, Google/Google naming it also reflects the importance of the algorithm.

The algorithm can calculate the value to reflect the relevance and importance of the web page.

PageRank determines the level of a page through the vast hyperlink relationship on the Internet, and interprets the link from page A to page B as page A voting for page B. Google/Google uses the level and voting target of page A or even the page linked to page A to determine the grade of B.

Simply put, a high-ranked page can boost the rank of other low-ranked pages.

The algorithm regards the entire Internet as a directed graph, web pages are nodes in the graph, and links between web pages are edges in the graph.

With the help of this algorithm, the degree of association of root search keywords of different web pages can be measured to rank the web pages.

For a long time, the search information you get when you type keywords in Google/Google.

The sorting order of the web pages corresponding to the series of web pages in the search information is sorted based on the PageRank algorithm.

The importance of this algorithm can be seen.

The Page Rank algorithm is not only used in the field of search engines.

It also crossed into the field of natural language processing (NLP).

The well-known TextRank algorithm in NLP is based on the PageRank algorithm.

The TextRank algorithm has always been the core algorithm of the extractive summarization algorithm.

Although the TextRank algorithm is currently mainly used in natural language processing.

But it does not mean that this algorithm cannot be applied to search.

After all, the TextRank algorithm and the PageRank algorithm used for search have the same root.

And Lin Hui made a generative text summarization algorithm (GTSA), although it seems to be a text processing algorithm on the surface.

But in fact, it also has the potential to play a role in the field of search in the future.

Compared with the PageRank algorithm, it crawls and sorts the web hyperlinks.

With the GTSA algorithm, Google can go a step further and directly grab the content of the top-ranked web pages under the PageRank algorithm to obtain corresponding information.

According to the criticality of information and search keywords, a secondary precise sorting is carried out.

This can undoubtedly greatly improve the accuracy of Google/Google searches.

Although it is difficult to ensure high search efficiency by nesting the generative text summarization algorithm under the PageRank algorithm with the current technology.

But who can guarantee that future servers and computing power will not advance by leaps and bounds?
What if the technology can skyrocket in the future?

And even if Google can't use the technology for search in a short time.

The powerful word processing ability shown by the generative text summarization algorithm is also a technology worthy of Google's attention.

Anyway, judging from the wording of the task issued by Google.

For the generative text summarization algorithm, Google not only pays attention to it.

And there is an urgent desire to acquire the technology.

……

However, as a specific executive, for the tasks proposed by Google:

——Evaluate the feasibility of the algorithm proposed by LIN HUI, and consider whether it can be reproduced in a short time according to the actual situation

Eve Carly was speechless.

Perhaps in the minds of those whose ass decides the head.

Once the technical route is clear, it is only a matter of time whether the technical reproduction can be realized or not.

But the truth is not that simple.

Anyway, Eve Carly, who tried all night, found it difficult to reproduce.

Regardless of the algorithm technology itself proposed by LINHUI.

It is the "LH text summarization accuracy measurement model" that LIN HUI took care of in the generative summarization algorithm patent

It would be difficult for other teams to build the same model from scratch.

Speaking of which, the construction process of the LH text summarization accuracy measurement model is very clear:
First, use language models to evaluate the fluency of algorithm-generated language;

Second, use a similarity model to assess the semantic relevance between text and abstract;
Third, in order to effectively evaluate the recurrence degree of entities and proper words, the original text information model is introduced to evaluate.

However, the goose is only very simple to say.

Speaking of putting an elephant in the refrigerator is also very simple, the same three steps:

- Open the refrigerator door, put the elephant in, close the refrigerator door.

Knowing how to do it is useless, the key is to implement it.

If there is no way to implement it, it is useless to have clear steps.

There are three steps in the construction process of the LH text summarization accuracy measurement model.

The first step is complicated.

How to build a language model?

Follow the technical route proposed by LIN HUI.

The language model modeling process includes dictionaries, corpus, model selection, etc.

The problem is corpus, which in linguistics means a large body of text.

Such texts are usually organized, with established formatting and markup.

Information related to the English corpus is relatively easy. After all, Eve’s team has a deep cooperative relationship with the linguistics of Oxford, Harvard, and Yale.

But it's hard to say how to deal with Chinese and other text prediction information.

Make bricks without straw.

It is useless to know the technical route without a corpus.

It is possible to abandon the research on Chinese and other news generative summaries for the time being.

But this is almost equivalent to giving up a huge market.

And the most important thing is that the algorithm proposed by LIN HUI can take into account both Chinese news summaries and English news summaries.

Then, will LIN HUI directly develop a function to process Chinese news into English summaries in the future?
It doesn't make sense for someone who can handle a text summarization algorithm to handle a translation algorithm, right? ?
The more Eve Carly thought about it, the more she felt that this possibility was very high.

Otherwise, why should a summarization software have an interactive style similar to that of translation software?
While they are still hesitating.

The opponent has already made great strides forward.

For a moment, Eve couldn't help feeling powerless after a fierce battle.

This is the so-called one step behind step by step behind it.

What is the assessment of this situation?

Just suggest to Google to buy back the algorithm of LIN HUI!
Although LIN HUI is from Huaguo.
But this kind of technology is not an important technology that involves the lifeline of the country.

It's impossible not to sell it!
If you don't sell it, you can also seek patent authorization!

It is not necessary to follow others to make wheels.

Eve didn't suffer/abuse/desire.

……

In fact, not only overseas are paying attention to the algorithm created by Lin Hui.

People in China have also noticed the movement made by Lin Hui.

It's just that the current domestic attention is mainly focused on the software level of Nanfeng APP rather than the algorithm level.

He Tianchang, vice president of Guoyang University and deputy/dean/director of the School of Computer Science, has been paying close attention to the software since the beginning of the Nanfeng APP.

Although it was late at night, he was still watching the downloading of the Nanfeng APP software through the third-party data network.

He Tianchang is not very famous in the outside world, but he is well-known in the software industry.

It is quite famous in the industry.It's not because of its tricks in software development.

But because of its old qualifications!All walks of life depend on qualifications, and the software industry is no exception.

However, He Tianchang's qualifications can also be seen. He and Wang Jmin were basically tinkering with software at the same time (96 years).

With this qualification, there are not many bigwigs in the Internet/software industry across the country who can compete with him.

It doesn't matter if you have this kind of qualification and don't do much in the software industry.

If you really want to attract investment, it is only a matter of making a few phone calls to casually attract tens of millions of investment in the Internet circle with your qualifications.

How did such a person who was originally engaged in software come to the university?

The story begins in the [-]s, after feeling the wave of the Internet.

He Tianchang, who just graduated from California Institute of Technology with a PhD in computer science, gave up a good job in the United States.

Returning to China to participate in entrepreneurship, when Wang Jianmin was tinkering with anti-virus software, He Tianchang also created several software.

However, it is different from the Jmin anti-virus software that the former fiddled with.

The software that He Tianchang tinkered with was generally niche but highly professional paid software.

Although this software is quite remarkable from the current point of view.

But such software was destined to be born at an untimely time in China in 96.

Not only was such software in 96 untimely, but free software became popular for more than ten years after that, and there was no market for paid applications.

(ps:...except for paid piracy, the early history of js)
Facts have also verified this point. In the six or seven years in the software industry, He Tianchang has achieved nothing but reaped some early Internet dividends.

More often than not, it is outside of experiencing the success of others.

Although the money he earned pales in comparison to people in the same period, his income is still enough to envy ordinary people.

But people's pursuits are different, what He Tianchang wants more is self-realization.

However, a person's destiny is not enough to struggle alone, and it is useless to toss if it does not meet the demands of the times.

After tossing around with a few highly specialized niche software but not receiving any rewards, the disheartened He Tianchang decided not to toss around, and just happened to be recommended by a friend to enter Guoyang University.

After comprehensive consideration, He Tianchang felt that Guoyang University was okay.

Directly recommended by a friend to join the National Central University as a teacher.

The tree moved the dead, the dead moved the living, and after that, He Tianchang got along smoothly, climbing all the way to a height that many people could not reach.

Although it is at a height that ordinary people cannot reach.

However, because of his early experience, He Tianchang kept his eyes on the domestic professional software market.

However, He Tianchang was very disappointed all the time, there were quite a few software with professional names.

There is no real professional software.

If a country's software industry can't produce the most advanced industrial software, it's fine, but professional software can't be produced?
He Tianchang has always been brooding about this.

But the advent of Nanfeng APP broke his knot.

Considering many factors, this software does have the quality that professional software should have.

And the professional mode of this software is in line with all the characteristics that professional software should have.

It can be said that He Tianchang was very pleased to see such software coming out in the domestic software market.

But he immediately became worried again, even if there is excellent professional software, wouldn't it be an inevitable fate if there is no market?
Looking at the third-party data of Nanfeng APP, although the download volume is good.

But the number of people paying for the professional model is only in double digits.

How can this work? People who make garbage games make a lot of money, but people who are serious about making professional software don't get rewarded?
What's the point? ? ?
no! ! !Such an excellent software developer cannot be allowed to repeat his past tragedies.

He Tianchang decided to do something, silently picked up his cell phone and dialed an old friend.

(End of this chapter)