What, how, and why: The dynamic of advanced NLG

Kristian Hammond offers an overview of advanced natural language generation (NLG), a subfield of artificial intelligence, and the assorted technical systems involved with this emerging technology, along with the mechanisms that drive them.

There is an immediacy to language. Unlike charts, graphs, or even the simplest of user interfaces, language provides a direct line of communication between a computer and a user. The other methods of communication require interpretation on the part of the recipient. Users have to figure out the meaning of a visualization or set of numbers in a way that they don’t with well-constructed language. This implies, however, that the burden of interpretation is on the shoulders of the systems that generate this clear and concise language. Systems that perform advanced natural language generation (NLG) need to be able to perform this interpretation of the data and map that interpretation onto understandable language. In particular, they need to attend to three elements: what needs to be said, how to go about saying it, and why it needs to be said. They have to go beyond the language and dive deeply into the data and the goals of the communication in order to get to an interpretation that supports the fluid language that we expect in human communication. Kristian Hammond offers an overview of advanced natural language generation and the assorted technical systems involved with this emerging technology, along with the mechanisms that drive them.