Advice on how to evaluate an NLG system by asking human subjects to rate the system, in an artificial experiment context (ie, not real world usage). This is the most common type of NLG evaluation
Evaluation in NLP/NLG has a long way to go before it reaches the standards of hypothesis testing in clinical medicine
Which is a better way to present information, text or graphics (information visualisation)?
Is it the case that important papers with major impact are *harder* to publish? This certainly seems to have been my experience
Perhaps not surprisingly, there are major differences in emphasis between commercial and academic work on NLG
I dislike talking about “NLG” vs “templates” because these terms are poorly defined. I prefer to talk about five levels of sophistication in generating texts, which I describe in this post
Simplenlg has evolved over 20 years from a set of Java classes I threw together for a project, into the most widely used open-source NLG package. In this blog I give a quick summary of its history.
Machine Learning and statistical corpus techniques are often very useful in Natural Language Generation, but they are not the best way to solve all NLG problems.
Ehud Reiter's Blog 