A text to understand natural language generation - NLG

Natural language generation-NLG is NLP An important part of his main goal is to reduce the communication gap between humans and machines, transforming non-verbal data into language formats that humans can understand.

In addition to introducing the basic concepts of NLG, this article will also introduce NLG's 3 Levels, 6 steps and 3 typical applications.

To learn more about NLP-related content, please visit the NLP topic, and a 59-page NLP document download is available for free.

Visit the NLP topic and download a 59-page free PDF

 

What is NLG?

NLG is part of NLP

NLP has 2 core tasks: NLU and NLG

NLP = NLU + NLG

Natural language generation-NLG is an important part of NLP. NLU is responsible for understanding content, and NLG is responsible for generating content.

Taking a smart speaker as an example, when the user says "What time is it?", first of all, you need to use NLU technology to judge the user's intention, understand what the user wants, and then use NLG technology to say "now 6 point 50 points."

 

Natural language generation-what is NLG?

NLG - Convert non-language format data into a human-readable language format

NLG is designed to bridge the gap between humans and machines and convert non-verbal data into human-readable language formats such as articles and reports.

Natural language generation-NLG has 2 ways:

  1. text – to – text: text to language generation
  2. data – to – text: data to language generation

Cases that turn data into language

 

XLGX Levels of NLG

XLGX Levels of NLG

Simple data merge:A simplified form of natural language processing that will allow data to be converted to text (via Excel-like functions). For the association,Mail merge(MS Word mailmerge), for example, where the gap is populated with data that is retrieved from another source, such as a table in MS Excel.

Simple data merge - mail merge

Templated NLG : This form of NLG uses the template driver mode to display the output. Take the football game score board as an example. The data remains dynamically changed and is generated by a predefined set of business rules, such as if / else loop statements.

Templated NLG

Advanced NLG : This form of natural language generation is just like humans. It understands intent, adds intelligence, considers context, and presents the results in insightful narratives that users can easily read and understand.

 

NLG's 6 steps

NLG's 6 steps

Step XNUMX: Content Determination – Content Determination

As a first step, the NLG system needs to decide which information should be included in the text being built and which should not. Usually the data contains more information than the final message.

 

Step XNUMX: Text Structuring-Text Structuring

After determining what information needs to be communicated, the NLG system needs a reasonable order of organizing the text. For example, when reporting a basketball game, it will give priority to "when", "what place", "what 2 team", then "outline" and finally "the ending of the game".

 

Step XNUMX: Sentence Aggregation-Sentence Aggregation

Not every piece of information needs a separate sentence to express. Combining multiple pieces of information into one sentence may be more fluid and easier to read.

 

Step XNUMX: Grammaticalization – Lexicalisation

Once the content of each sentence is determined, it can be organized into natural language. This step adds some conjunctions between the various information and looks more like a complete sentence.

 

Step XNUMX: Reference Expression Generation – Referring Expression Generation|REG

This step is very similar to grammaticalization. It is to select some words and phrases to form a complete sentence. But the essential difference between him and grammaticalization is that "REG needs to identify the domain of the content and then use the vocabulary of the domain (rather than other domains)."

 

Step XNUMX: Language Realization – Linguistic Realisation

Finally, when all relevant words and phrases have been determined, they need to be combined to form a well-formed complete sentence.

 

NLG's 3 typical application

Regardless of how NLG is applied, most of them are for the following 3 categories:

  1. Ability to generate personalized content on a large scale
  2. Help humans gain insight into data and make data easier to understand
  3. Accelerate content production

 

Here are some typical applications for you:

NLG's 3 typical application

Write news automatically

News in certain areas has obvious rules, such as sports news. A lot of news has been done with NLG.

'Tencent robots have written thousands of articles per day. The news you read may be written by AI."

 

Chat robot

Everyone knows that chat bots start from Siri, and in recent years there has been a boom in smart speakers.

In addition to the areas that are familiar in everyday life, customer service is being replaced by robots, and even some telephone customer service is also a robot.

'The customer service that talks to you is a robot!"

HomePod has built-in chat siri

 

BI interpretation and report generation

Almost every industry has its own data statistics and analysis tools. These tools can produce a wide variety of charts, but the output of conclusions and opinions still depends on people. A very important application of NLG is to interpret these data and automatically output conclusions and opinions. (As shown below)

NLG automatically generates a report of data interpretation

 

Final Thoughts

Natural language generation-NLG is an important part of NLP. Its main purpose is to reduce the communication gap between humans and machines, and to convert data in non-verbal format into a language format that humans can understand.

 

NLG's 3 levels:

  1. Simple data merge
  2. Modular NLG
  3. Advanced NLG

 

NLG's 6 steps:

  1. Content Determination – Content Determination
  2. Text structure – Text Structuring
  3. Sentence Aggregation – Sentence Aggregation
  4. Grammaticalization – Lexicalisation
  5. Reference Expression Generation – Referring Expression Generation|REG
  6. Language Realization – Linguistic Realisation

 

3 purposes for NLG applications:

  1. Ability to generate personalized content on a large scale
  2. Help humans gain insight into data and make data easier to understand
  3. Accelerate content production

 

NLG's 3 typical applications :

  1. Write news automatically
  2. Chat robot
  3. BI interpretation and report generation

 

Baidu Encyclopedia Version + Wikipedia

Baidu Encyclopedia version

Natural language generation is the study of the ability to make computers have the same expression and writing as human beings. That is, according to some key information and its expression in the machine, through a planning process, a high-quality natural language text is automatically generated.

Natural language processing includes natural language understanding and natural language generation. Natural language generation is a branch of artificial intelligence and computational linguistics. The corresponding language generation system is a computer model based on language information processing. Its working process is contrary to natural language analysis. It starts from the abstract concept level and selects and executes certain Semantic and grammatical rules to generate text.

Read More

 

Wikipedia version

Natural Language Generation (NLG) is an aspect of language technology that focuses on generating natural language from structured data or structured representations such as knowledge bases or logical forms. When this formal representation is interpreted as a model of psychological representation, psycholinguists prefer the term language generation.

It can be said that an NLG system is like a translator that converts data into a natural language representation. However, due to the inherent expressiveness of natural language, the method of producing the final language is different from the method of the compiler. NLG has been around for a long time, but commercial NLG technology has only recently been widely used.

Read More