Time Travel: 2014

Although in this time and space, Western academic circles have devoted a lot of effort to the study of text summarization.

However, the research progress of text summarization in the West in this time and space is still somewhat different from the research status of text summarization in the West in the previous time and space.

Although objectively speaking the difference is not that big.

But comprehensively speaking, the research on text summaries in the Western world in this time and space is two years slower than that in the previous time and space.

(As for China, it goes without saying that the academic community at this time is still accustomed to crossing the river by feeling the falcon's sauce.

To be honest, this approach is not entirely wrong and can avoid wasting resources.

But it would be too passive to always be like this.

If you want to be the boss, you must dare to be the first in the world)

Although the research in related fields in this time and space is only two years slower in pace.

But two years is enough time to change a lot of things.

What's more, Lin Hui already had seven years of information advantage.

Under the circumstances, Lin Hui had an information advantage of nearly ten years.

Some people may wonder how Lin Hui can fully utilize his information advantage after only working for three years.

Although Lin Hui only worked for three years in his previous life, it is not an exaggeration to say that he has six years of work experience.

As for where did the extra three years of work experience come from?

It’s too much to say that it’s just tears, it’s the result of working overtime.

I have to say that these are all “blessings”.

Working overtime is a blessing.

How could Lin Hui have a chance to be reborn if he didn't work crazy overtime like this?

Even if there is a chance to be reborn.

How could Lin Hui remember those boring things so deeply if he didn't work overtime like crazy?

But these are things of the past.

Because of various experiences in the past, Lin Hui is a well-deserved strong man in this time and space.

As for other researchers in the same field, Lin Hui respects their efforts.

But I have to say: Sorry, you are really weak!

It wasn't that Lin Hui was just talking nonsense.

Lin Hui previously worked on all the technologies involved in the generative text summary algorithm.

If it is completely understood by the research team in this time and space.

At least it can accelerate the research progress of natural language processing and neural network learning around the world in this time and space by nearly a year.

Of course, this means that if you understand it immediately, it can speed up the time by nearly a year.

If it takes these research teams two or three years to make corresponding progress, it will actually hinder their normal progress.

Put aside the patent of generative text summarization.

Just the LH text summary accuracy measurement model that Lin Hui came up with when he was working on generative text summarization is awesome enough.

If this technology can be mastered by the research team of this time and space, it will also help their research.

Although Lin Hui had clearly stated how to build the model at the beginning, he still needed step-by-step instructions.

(To build a model, you must first use a language model to evaluate the fluency of the language generated by the algorithm, then use a similarity model to evaluate the semantic correlation between the text and the abstract, and finally, in order to effectively evaluate the degree of recurrence of entities and proprietary words, introduce the original text information model to evaluate)

But researchers at this time still seemed curious about how Lin Hui constructed this measure.

Lin Hui remembered that Eve Carly had expressed confusion about how the "LH text summary accuracy measurement model" was constructed in the email he sent earlier.

Lin Hui remembered that Eve Carly was not only curious about how Lin Hui solved the corpus problem.

The confusion mainly focuses on what method Lin Hui used to construct the similarity model.

Lin Hui was surprised when he learned that researchers at research institutions affiliated with the world's top universities were actually curious about this matter.

Lin Hui built a "gorgeous building" with great ambition.

I originally thought that people in this time and space would be curious about how Lin Hui built this building.

Unexpectedly, I was first asked how the materials used to build the building were made.

This was Lin Hui's intuitive feeling when he received Eve Carly's email.

However, if it is as Eve Carly introduced in the email, Lin Hui can also understand why Eve Carly is confused.

The architecture involving similarity models is generally calculated.

The semantic similarity of the two texts is measured by calculating the semantic text similarity.

Generally speaking, the smaller the semantic similarity value, the greater the semantic difference between two texts, and the lower their similarity at the semantic level;

On the contrary, the larger the value, the more similar the semantics expressed by the two texts are.

Perhaps in people's eyes, distinguishing similar texts is a very simple thing?

Can't this be done by just reading it?

But you must know that distinguishing similar texts does not require humans to distinguish, but requires machines to distinguish similar texts.

It is indeed not easy to build a similarity model. After all, human language expression is extremely complex.

Not to mention that in most professional articles, there are many synonyms, abbreviations, specific words and changeable syntactic structures in the text.

These greatly increase the difficulty of calculating text semantic similarity.

But this problem cannot be solved unless it is solved. Lin Hui knows that calculating text semantic similarity is a very important branch field.

In the field of information retrieval, semantic text similarity calculation plays an important role in tasks such as text classification, text clustering, and entity disambiguation;

In the field of artificial intelligence, semantic text similarity algorithms are also needed to support tasks such as question answering systems and intelligent retrieval.

In addition, semantic text similarity calculation is also widely used in natural language processing tasks, such as plagiarism detection, text summarization and machine translation.

In short, the research on similarity models represented by semantic text similarity algorithms has important application value.

If we do not solve the problem of calculating text semantic similarity, let alone how to further text processing.

Put aside the problem of asking machines to distinguish between similar texts.

Just getting a machine to recognize text is extremely difficult.

Natural language generally refers to language that humans can understand. For example, the text you see is natural language.

But when we need machines or computers to process natural language.

Machines/computers cannot directly understand these symbols (Chinese characters, letters, punctuation marks, etc.).

These symbols must first be digitized before they can be entered into a computer for subsequent processing.

Just digitizing it is of little use.

Some other content must be introduced to reflect the attributes of the word.

Just like we can't know from an ordinary code name whether this string of numbers represents subscription, collection or reward.

In short, it is impossible to tell the attributes corresponding to each string of numbers from just a code name.

This issue is also one of the hot research topics in calculating text semantic similarity.