On this page there are pointers to several online NLG systems. These systems range across several "domains" of application. The last one, TEMSIS, is a multilingual generation system (MLG); the others are for English. You can try these out: try and get a sense of how the texts produce vary depending on the input that you give. What kind of input must you give? What kind of knowledge must have been encoded in the NLG program in order for it to get from the input that you give to the text that it produces? For the MLG system, you can consider what the differences or similarities would be when compared with a Machine Translation system: do the two perform the same?
Medical domain:
Some more background information about why one might use NLG for medical information is given in a paper that can be downloaded here: Bental, D., Cawsey, A.J. and Jones, R.B. (1999) Patient Information Systems That Tailor to the Individual (word file, .5MB). Journal of Patient Education and Counselling, 36:171-180
Museums domain:
Some successive texts generated with ILEX are given here along with the reader's actions between them: try to see where there are dependencies, how do the texts differ from texts that just pre-written by a museum website designer? How could they be made more responsive still?
Environmental information:
General information: