Hi,
I don’t think that this is feasible. To generate correct sentences you need a lot of labeled training data. Since in your table every row contains similar information, the way to go is using a template. I still don’t really see what you would expect from the NLP approach. How do you imagine it should work? How should it know what words to use to fill the gaps between the terms in your table and how do you make sure it does not always use the same sentence structure? Without some more information I think you won’t get any help here.
Kind regards
Alexander