NLG texts must be correct pragmatically as well as semantically. In particular, they must not contain statements which are contextually misleading even if they are literally true. We badly need better techniques for evaluating pragmatic accuracy as well as generating pragmatically correct texts.
There is a lot of uninformed criticism of rule-based NLG in academic papers. In this blog I explain at a very high level how such systems work and what some of the main challenges are in building them.
The fundamental challenges of building useful data-to-text NLG systems are the same regardless of whether we build systems with rules or transformers. We need to understand where NLG is useful, choose good content to communicate, robustly deal with edge cases, allow users to configure and control the system, and evaluate properly. I’d like to see more research on these fundamental issues, regardless of technology used.
One of the highlights of INLG for me was the panel on “What users want from real world NLG”. I summarise a *few* of the really interesting points made about trust, authoring, configurability, human-in-loop, and other key issues for real-world NLG users.
We’ve just completed a shared task on evaluating accuracy of NLG texts. This was really interesting, and amongst other things showed that current neural data-to-text systems struggle to learn how to use some words which have clear but relatively complex definitions.
Ehud Reiter's Blog 