Skip to main content

What is Natural Language Generation?

Since the beginning of 2017, the term Natural Language Generation has become more and more common. But its rise in popularity also led to some confusion. Even the Forbes article that declared NLG one of the top trends for 2017 listed several companies that do Natural Language Processing, not Natural Language Generation! How can you blame them? NLU, NLP, and NLG are all different subsections of AI, which can be very confusing. So, what is Natural Language Generation exactly?


What Is Natural Language Generation: What It Does & Doesn’t Do

Natural Language Generation (NLG) is a subsection of Natural Language Processing (NLP). NLG software turns structured data into written narrative, writing like a human being but at the speed of thousands of pages per second. NLG makes data universally understandable and seeks to automate the writing of data-driven narratives like financial reports, product descriptions, meeting memos, and more.

Importantly, while NLG software can write, it can’t “read”. NLG turns structured data into human language but it is not able to, for example, read a news story and pull figures out of it. In fact, the science of taking unstructured data like a book and making it structured is called Natural Language Understanding.


Learn the differences between Natural Language Understanding and Natural Language Generation. Download this Infographic.


What Is Natural Language Generation: How It Works

Like any subsection of computer software, the functionalities of what software vendors call “NLG software” varies, with some even trying to commercialize simple templating systems which generate narrative like “mad libs” (yes, I know I am dating myself with that reference). Basically, these simple NLG systems have prewritten sentences with holes in them. For example, “The weather in <<location>> is <<temp>> degrees”. These simple systems turn one data point into a sentence which works for some very use cases but aren’t scalable, easily maintainable or generally enterprise ready.

Next-generation NLG is software able to summarize larger amounts of data and able to explain WHY numbers are what they are. These systems allow companies to generate both back-office and front-office reports since they write more than just descriptive narrative. In fact, they go further with tools that are able to explain analysis, like Yseop Compose, which writes in multiple languages and installs on a private cloud or on premise.


How Is Natural Language Generation Used?

NLG software is used to automate the writing of written reports, but more broadly it fits into the data to data-driven decision-making workflow. NLG is really the last mile in this process. Data collection and analysis have already generally been automated, it’s the last step of explaining the results of the analysis in plain English (of whatever language) where NLG helps.


Download this eBook for more detail at how NLG fits into the data-drive decision-making workflow.


Now when someone asks “what is Natural Language Generation”, hopefully, you will have an answer. NLG is software that writes like a human being from structured data, generating narrative at the speed of thousands of pages per second.