I have seen some examples on the web, showing how to give the feeling to have a semi-transparent volume.
I have seen this with subsurface scattering, volumetric scattering and volumetric absorption.
I don’t know exactly what are their main separated purposes and their differences.
I have seen how they’re useful to make the human skin semi-transparent over the ears, or a gummy cheese, or a transparent chess made of glass, ecc. But I want to understand more about their most common usages and the differences. 
Indeed both subsurface scattering and volumetric scattering simulate light passing through a volume. I think you can have a similar result using volumetric scattering rather than subsurface scaterring (SSS) . But it then would be slower to compute because SSS do some approximation and is made for material that are mostly opaque , but with a little translucency that let light scatter inside, like wax, milk, skin ect…
Volumetric scattering simulate the same effect but it’s a volume shader, the main use is to render smoke simulations or fog. With material that can vary a lot in transparency .
SSS is a surface shader , so it won’t compute things that are inside the mesh, but it simulate the scattering effect…
Hope all that help to make a little sense, it’s a question a bit difficult to answer clearly for me.
SSS is less robust than true volumetric scattering, but it’s far easier to do things like simulate varying levels of scattering (in terms of depth and color). This is because the differentiation can be done with standard textures and/or vertex coloring (not so with true volumes which only take 3D textures as input).
What makes it less robust is primarily the inability for light to actually continue beyond an SSS-shaded mesh (illuminating surfaces beyond thinner areas for instance). You can combat this though by mixing in the use of the translucent or refraction node (the latter having the roughness setting above 0), in turn having their visibility modulated by use of the ray depth output in the lightpath node.