NLG approach can be used to automatically generate texts through natural language imitation. After machine translation, it is one of the main goals of Artifical Intelligence research since 1950s. However, text generation that imitates human language was born in the 1970s. At that time, machine’s answers are not pre-recorded segments anymore but true generated answers.

Nowadays, the number of textual data has exponentially increased with the boom of the Internet and Big Data. Companies want more than answers to specific questions (with chatbots, for example or automatic answers to emails), they need the ability to generate longer texts in a complete automatic way and for several application fields.

NLG is particularly used in marketing or customer relationship. Some real estate companies use this technology to generate accomodation advertisements, for example. Indeed, text generation must be programmed for a specific goal. Writing style, syntax, terminology should be adapted in accordance with the type of text, as the universal text generator is not yet ready to be invented.

In order to create a relevant and efficient text generator, its specific goal must be established (answering customer requests, generating marketing texts, answering questions on forums, etc.) in order to specialise the robot and train it for a specific task. A single-task robot (that is to say a robot that is specialised) will be much more efficient than a general robot.

Besides the obvious advantage to transfer the writing tasks and other time-consuming and repetitive tasks to the machine, NLG allows a better SEO. The investment can be really interesting for companies, regardless of the industry, and NLG can be involved in some unexpected ways in some fields.