For our website, we have started out using (as with all projects) with sentence segmentation. For reasons of SEO, we have to change that to paragraph segmentation. By translation on paragraph level, we can better direct commercial messages to the correct audience.
OmegaT might be able to generate translations for a new (paragraph) source segment, based on the concatenation of the original source segments, by concatenating the original target segments into a new target segment.
Such a change would help our translation process, similar to the already existing functionality which tries to guess the TMX contents when going from paragraph to sentence level segmentation.