Natural Language Generation (NLG) is a technology for building software systems that can “write”, ie, that can produce explanations, summaries, narratives, and so forth in English and other human languages.
For a (relatively) non-technical description of NLG, see the relevant wikipedia pages (I wrote most of these). You can also read my book, Building Natural Language Generation Systems. It is pretty old (published in 2000), but I think its still a reasonable description of the fundamental concepts of NLG.
Natural Language Generation is a type of Natural Language Processing (NLP). Other well known types of NLP include machine translation, sentiment analysis, and speech recognition. NLG and NLP are types of Artificial Intelligence (AI), along with topics such as robotics, computer vision, and game playing.
The academic NLG community has yearly conferences where the latest NLG findings are presented. These conferences are organised by ACL SIGGEN, check its web pages for information on upcoming events. Papers presented at these events are available online through the ACL Anthology.
There is growing commercial interest in NLG, much of which focuses on data-to-text, that is systems which combine NLG and data analytics in order to generate summaries, explanations, etc of structured data such as time series. I spend part of my time working for Arria NLG, a commercial NLG company which does a lot of data-to-text work,