Time Travel: 2014

After hanging up the phone call from Huang Jing.

Lin Hui began to review the professional mode he had previously launched on the Nanfeng APP.

This feature of the professional mode of Nanfeng APP might have been scolded in previous lives.

Lin Hui could even guess how people would question:

——I’m crazy about money. The professional version only has this function?

He might even ask Lin Hui if his conscience doesn’t hurt?

Sorry, Lin Hui's conscience doesn't hurt at all.

I even wanted to laugh a little bit.

The so-called news summary technology will indeed be useless in the next few years.

The third, fourth, and even sixth generation summary algorithms are not worth mentioning in the next few years.

After the deep integration of artificial intelligence in later generations, it is no exaggeration to say that technological progress in natural language processing has been rapid.

In a few years, the fees for news summary processing will be completely cheap.

Taking 2021 as an example, how does a large Internet company charge for processing news summaries?

For medium and large enterprises that require flexible payment, the news summary algorithm program provided by this Internet company can be called free of charge 50,000 times a day.

The number of paid calls is unlimited.

Even paid calls are not very expensive, basically only about 2 yuan per thousand times.

The QPS configuration is 20 concurrency.

The so-called QPS refers to the query rate per second (QPS, Queries-per-second).

It is a measure of how much traffic a specific query server handles within a specified period of time.

On the Internet, the performance of a machine that serves as a domain name system server is often measured by query rate per second.

It can be said that 20 concurrent QPS can fully meet the needs of medium and large enterprises.

For large customers who require special models, they can even enjoy exclusive prices.

Not only that, but also supports privatized deployment for such large customers.

Let’s leave aside these professional enterprise users and only talk about individual users!

When individual developers choose the 2~5QPS quota, their daily free usage can actually be as high as 500,000 times.

It can be said that the cost of news summary processing in the next few years will be ridiculously low.

But does it make sense to compare prices in the next few years to current charges?

It makes no sense at all.

Lin Hui also knows that the algorithm he is tinkering with now will not be anything special in the next few years.

But Lin Hui also knew that the algorithm he used in his current software was Niubi.

Technology is one thing, and two things are two.

To be ahead is to be ahead, to be invincible is to be invincible.

The algorithm used in Lin Hui Nanfeng APP is amazing.

What does it mean to be ahead?

It means that other technical teams must follow the path you have taken if they want to move forward.

If other technical development teams want to implement the algorithms used in Linhui software, they have to follow the technical route honestly.

Issues involving text classification, language modeling, document information screening, etc. of generative text summarization

How can other technical teams know why without doing it all over again?

Technical research is not pie-in-the-sky.

Not to mention knowing the reason, it is a bit difficult for other technical teams to know the reason now.

After all, some of the concepts involved in some of Lin Hui's generative summary algorithms are groundbreaking.

For example, generative text summarization involves sentiment analysis of natural language processing.

After all, generative text summarization will involve the emergence of some new words or phrases.

The emotions expressed by these words or phrases must be consistent with the original news content.

This concept is basically not covered by teams that have previously studied extractive text summarization.

The vocabulary involved in the extractive text is all vocabulary in the news.

Unless it's weird news.

Otherwise, the emotional color of the words extracted from the news is basically consistent with the emotional color expressed in the news.

(ps:... BBC, CNN cried and fainted in the toilet)

Therefore, the algorithm team of extractive text summarization basically did not encounter the research direction of NLP sentiment analysis when studying extractive summarization before.

But now these teams that previously studied extractive text summarization algorithms want to get involved in generative summarization algorithms.

It is necessary to re-engage NLP sentiment analysis.

And NLP sentiment analysis is just one of Lin Hui’s many small contributions.

However, for this small contribution, other technical teams have to build bridges across mountains and rivers if they want to reproduce it.

And this journey will definitely require a lot of effort and money.

Money comes second. Internet companies have plenty of money.

Time is the most precious wealth.

This is why Lin Hui believes that generative text summarization can make him tens of millions of dollars in wealth.

As for Nanfeng APP professional version users.

With less than 20,000 yuan, you can experience the algorithm that has eluded countless top NLP teams.

Shouldn’t users be grateful?

(ps:...funny saves life, please put this sentence in the face of a big factory)

And Lin Hui’s charges are cheaper than similar software?

The most important thing is that Lin Hui also reserved a professional cooperation contact number in the description of the professional mode options.

If you are not satisfied with the price, you can definitely negotiate.

As for complaining about Lin Hui’s expensive price under such circumstances?

It is probably because there is no demand for the professional mode of this software at all.

Isn’t there a similar scene in The Analects of Confucius?

"The beauty of Tang Di is the opposite. Don't you think about it? The house is far away."

Confucius said: "If you don't think about it yet, how far away is it?"

Anyway, Lin Hui didn't think there was anything wrong with his thinking.

Users who do not need the professional version can definitely experience the ordinary version.

There is no need to go far and wide to pursue the professional version.

Compared with the professional version, the normal version only has two minor functional differences.

It does not affect the user experience of non-professional users at all.

In addition, a slightly higher price can help users of the professional version re-evaluate whether they really need the professional mode?

All in all, Lin Hui is satisfied with the pricing strategy he has set.

It can be said that it is very considerate for users.