Differential compression is all about describing a dataset B based on another data set A as well as the changes performed on A in order to reach B.
You can see differential compression as a function Delta = D(A,B) where Delta is the difference or the description of the changes.
The same way you can perform the opposite operation where B = R(A,Delta).
Assuming that Delta can be represented in a more compact form than B, this is a compressed way of storing B assuming you also store the earlier version of the data A.
The demo project related to this subject can be found here.
The full text, which is in Danish, is in the form of a PDF document that can be downloaded at the bottom of this page.