Time Travel: 2014

Lin Hui is not only very easy-going, but also never gives people a sense of distance.

After meeting Lin Hui, Hu Xin seemed to understand the saying "talented people have temper, while talented people have no temper".

Judging from this perspective, Lin Hui is undoubtedly a talented person.

Lin Hui didn't know what Hu Xin was thinking deep inside.

In fact, treating others with respect is just a subtle influence from the surroundings.

Although Lin Hui is a time traveler, he is not an orphan and does not lack tutoring.

It's certainly not unusual for you to have a quick chat with someone you just worked for.

What's more, these people are all reporters.

And he is the best reporter.

It is no exaggeration to call these reporters the best.

Although the newspaper "Digital Youth" sounds a lot like some ordinary electronic magazines.

But what Lin Hui knew was that the publication "Digital Youth" was one of the sub-publications of "China Youth Daily".

In other words, many of the reporters who interviewed Lin Hui this time were reporters from China Youth Daily.

"China Youth Daily" is a serious newspaper directly under the tutelage.

Although its status in official newspapers and periodicals is lower than that of official newspapers and periodicals such as People's Daily and Guangming Daily, as well as the military's military newspapers.

But its status is also quite high.

To be able to serve as a reporter in such a transcendent newspaper must be of a very high standard.

What is a reporter?

Those are all Lin Hui's customers.

Why are these people called customers?

You must know that although the core algorithm of Nanfeng APP, generative text summarization, this algorithm can basically cover all text summary processing scenarios.

But when Lin Hui first started working on Nanfeng APP, he simply wanted to build a news summary software to benchmark Yahoo News Summary software.

In other words, Lin Hui's original intention was to process news, and the text processing of other scenes was largely just a matter of hunting down rabbits.

When Lin Hui focuses on news processing, reporters are naturally Lin Hui's customers or potential users.

At this time, these outstanding reporters were high-quality customers in Lin Hui's eyes.

Nanfeng APP has been online for some time, and many copies of the professional mode have been sold.

But Lin Hui remembers that when the software Nanfeng APP was first launched, it was actually very hasty.

At that time, Lin Hui only conducted a rough market survey and did not conduct too in-depth market research.

Even the pricing of the professional model is largely based on Yahoo's guidance.

Even after Lin Hui officially launched the Nanfeng APP, he paid close attention to the users who commented on the Nanfeng APP.

Because we adopted the comments from users below in the comment area of ​​Nanfeng APP.

Lin Hui has created a professional mode based on Nanfeng APP that is more in line with the usage habits of professionals.

In this version, Lin Hui has canceled the input limit of 10,000 Chinese characters for news in professional mode.

But this is still not enough.

Lin Hui also needs to listen to professional feedback from professionals.

After all, only professionals know what they really need.

There are some excellent journalists like this coming to your doorstep.

Moreover, before the interview, Lin Hui also noticed that among the reporters interviewing Lin Hui, there were also journalists who specialized in copy editing.

How could Lin Hui miss such an opportunity?

Through some conversation, Lin Hui learned that the copy editor accompanying Hu Xin this time was named Qiu Jiachun.

A slightly shy girl who looks younger than Hu Xin, probably in her twenties.

Although he looks a little shy, he is a journalist after all.

When Lin Hui asked her for advice, she still expressed her opinion.

She suggested that Lin Hui introduce the ability to process multiple news in the professional mode.

To be honest, this is not the first time Lin Hui has received this suggestion.

Lin Hui has heard more than one user make suggestions in this regard before.

At that time, Lin Hui saw in the comment area of ​​Nanfeng APP that many people suggested that the developer add the function of processing multiple news.

But Lin Hui remembers that there were big differences among the people in the comment area on whether to use the function of processing multiple news, and there were even fights in the distance.

Proponents argue that the ability to multitask increases productivity;

Opponents believe that multitasking can easily lead to distraction and affect concentration at work.

At that time, the software's comment section was very open, and it was also limited by technical issues.

Lin Hui's original approach was to maintain the status quo on this controversial point.

After all, there is no need to worry about multi-tasking. Even if he wants to engage in professional mode, Lin Hui's workload is much less.

I originally thought that this matter would be over, but I didn't expect that now I would actually come into contact with the outstanding customers that Lin Hui talked about.

The first piece of advice I heard was to add multitasking.

This is more embarrassing.

However, with Lin Hui’s character of doing something perfect if he doesn’t do it, it’s impossible to avoid it easily.

Since Lin Hui has asked the question, we must find out why journalists who are mainly engaged in copy editing work need to multi-task.

According to Qiu Jiachun, when copy editors deal with original news manuscripts, they often do not just deal with one person's interview materials.

Instead, we need to process many similar news.

For example, an event in a youth competition.

When visiting an event, I often interview many different contestants.

During actual processing, the speeches of these contestants must be processed separately.

This process often takes a long time if the summary does not have a multi-tasking mechanism.

Therefore, Qiu Jiachun suggested to Lin Hui to add the multi-tasking function.

As for the differences among users on whether to add a multitasking mode.

Qiu Jiachun felt that if Lin Hui had difficulty making a choice about this matter.

Might as well simply leave the decision-making power to the user.

Turn on the multi-tasking mode switch in the professional mode of Nanfeng APP.

Users who think multitasking affects concentration can choose not to turn on the corresponding multitasking mode.

Users who need multitasking can choose to turn on the corresponding multitasking mode.

He listened carefully to Qiu Jiachun's appeal.

Lin Hui learned that Qiu Jiachun's demands were basically focused on multitasking.

In fact, it is not troublesome to introduce multitasking into Nanfeng APP.

Even Lin Hui was able to make Nanfeng APP a hundred times more efficient than Qiu Jiachun imagined.

But in many cases, technological development cannot but consider its impact on society.

If a technology advances rapidly overnight, it will cause many social problems.

For example, if software such as news summary is really efficient, it will increase hundreds of times and thousands of times overnight.

Many news staff like Qiu Jiachun who are purely text-based are likely to lose their jobs immediately.

In addition, if many things are to be moved out, the rationality of the transportation must be considered.

In terms of the long-term development of generative text summarization algorithms, Lin Hui has made relatively long-term plans and put them into practical actions.

But these are still not enough.

Even if the technology is advanced for a while, if it rests on its laurels, it will eventually be overtaken by its competitors.

In short, be prepared for danger in times of peace.

Relying on the patents declared by Lin Hui himself and acquired.

Lin Hui currently has a head start in world text summarization.

Even if Lin Hui has the upper hand, he can only reasonably develop (reasonably transfer) the generative text summarization technology to the fifth generation.

It is only the fifth generation of generative text summarization technology that still has limitations in many application scenarios.

Take the news/text summary multitasking aspect.

The fifth generation of generative text summarization is not well qualified for this application.

In fact, it is not only the fifth-generation generative text summarization algorithm that has trouble processing multi-task news summarization.

It is not easy for future generative summarization algorithms to handle multi-task news summarization.

It is almost impossible to achieve multi-tasking news processing relying on purely generative text summarization algorithms.

Even in the technical field, it is rare for one trick to become popular all over the world.

If you want to achieve efficient multi-tasking of news.

Maybe we have to wait until artificial intelligence matures.

The maturity of artificial intelligence will drive many fields that previously encountered bottlenecks to take off.

This includes text summarization.

When artificial intelligence matures, natural language processing projects such as text summarization will not only give rise to new possibilities.

Moreover, the cost for users in the field of natural language processing will also fall rapidly.

The future is bright, but the road is tortuous.

It will obviously take a long time to realize this wonderful prospect.

Especially in this time and space, many advances in machine learning are not as rapid as in the previous time and space.

There is no concept of machine learning now, but deep learning is still far behind.

And without deep learning in the true sense, there is no artificial intelligence in the true sense.

To some extent, deep learning is a subset of machine learning.

It is currently unrealistic to use artificial intelligence to promote multi-tasking progress in text summarization.

After all, Lin Hui still needs a long period of time to carry out some infrastructure construction in deep learning theory.

As for now, if you want to achieve multi-task text summarization.

The only more feasible method is to use big data to train news summary technology.

Through long-term training, the generative summary algorithm can handle multi-tasking of news.

Why does training generative summary algorithms involve big data?

In fact, algorithms are not castles in the air, and generally require long-term exploration in the practice of production and life.

The birth of an excellent model and algorithm is indispensable for continuous debugging and exploration.

Usually, debugging is inseparable from the support of data.

The reason why is usually the case.

This is because there will be exceptions in the model/algorithm debugging process involving some bosses with super intuitive intuition.

These bosses sometimes do not rely on data, but rely on intuition to micro-manage to a very efficient state.

But this does not negate the importance of data.

The boss's intuition should largely be based on experience gained from long-term data processing.

In short, data is quite important.

The core of big data is to utilize the value of data.

Proper use of big data and some research rules of big data in the algorithm adjustment process can often achieve twice the result with half the effort.

If a machine learning algorithm in natural language processing like generative text summarization is to achieve rapid development, it cannot do without the help of big data.

Although using big data to tune generative text summarization can only be said to be a simple application of big data.

However, relying on big data to train and adjust algorithms also takes time and cannot be achieved overnight.

Lin Hui estimated that he would follow the normal scientific research progress.

It will take at least a month or two to develop a summary algorithm that can handle multi-tasking news using big data.

Although there is a high probability that Lin Hui’s past life information contains ready-made algorithms that are perfectly capable of multi-task news summarization.

But Lin Hui thought it would be better to move things scientifically.

In short, moving is impossible in the near future.

Lin Hui has made too much noise recently.

If you get involved in big data in a short period of time, you may not know what problems will arise.

Although even if the forest ash is carried as usual, no one may question the level of the forest ash's technology.

After all, Lin Hui is far ahead in text summarization.

Many pursuers may already be in a state of numbness where they cannot find North.

Under this situation, even if Lin Hui takes action, these people may not be able to catch up with Lin Hui.

But as the saying goes, don't see what's hidden, don't show what's subtle, so a gentleman should be careful about his independence.

The opponent's numbness is not the reason for Lin Hui to act recklessly.

There is only a difference between 0 times and countless times for some things.

A move that did not follow the logic of the normal technological evolution sequence gave Lin Hui a taste of the sweetness.

Lin Hui cannot guarantee that he will not continue to abide by the laws of objective facts in the future.

After all, the temptation of fried fish in the fish pond is quite attractive.

But this "coolness" comes with the price of danger.

It is very dangerous to cause trouble and trouble if you do not abide by objective laws.

Perhaps one or two illogical moves will not attract attention.

But if things continue like this, it will inevitably not turn over.

I often walk by the river without getting my shoes wet.

To be on the safe side, it is better to follow objective logic.

A steady flow of water is the way to go.

It would be bad if the car overturns accidentally.

In short, even Lin Hui's previous enterprise-class hard drive must have a well-tuned algorithm that is perfectly capable of processing multiple messages.

Lin Hui didn't bother to carry out such an algorithm.

At least Lin Hui didn't have such thoughts in a short period of time.

What Lin Hui has to do now in terms of academics is to delve deeper into the direction of generative summary algorithms.

Although this direction of generative summarization algorithms seems not worth mentioning.

But behind it lies the vast world of natural language processing.

From this grand research field, Lin Hui believes that he can reach quite high academic heights.

The recent nomination for the second prize of the National Science and Technology Progress Award in this direction also illustrates the correctness of this direction.

In short, Lin Hui had no doubts about his choice of route.

Lin Hui had already thought a lot about this before.

Lin Hui didn't feel that he had made any omissions.

Once again, Lin Hui has more information resources than others.

Based on the decision made based on this information, Lin Hui believed that the direction he chose was correct.