If you're familiar with the cooking recipe instruction, 'season to taste'; that pretty much describes sharpening.
Current conventional thinking is that sharpening is to be performed in 3 stages.
1. Capture sharpening. Most digital cameras contain a 'low-pass' optical filter in front of the sensor to reduce aliasing at the sensor sites. This has the unwanted side effect of 'filtering out' sharp edges. Post processing software needs to restore this information, which can sort of be determined mathematically from the sensor matrix data. Lr has this built into the 'camera calibration profiles'. That's why the sharpening setting in Lr defaults to 25, so you can reduce it if you choose to.
2. Creative sharpening This is the 'season to taste" part, Lr provides both global and local tools to allow you to adjust the image to your sensibilities.
3. Output sharpening This is another round to adapt the image to the idiosyncrasies of the particular output mechanism. Display screens are different from printers, and matte paper is different from glossy, and yada, yada. Like capture sharpening, this is pretty well characterized, and deterministic. Once you know what the output channel will be, Lr can automatically match it.
Most of this derives from a seminal paper authored by the late Bruce Fraser and Jeff Schewe in conjunction with the rest of the folks at Pixel Genius who developed sharpening software . A google search will turn it up. The good news is the Pixel Genius technology forms the basis of Lr's sharpening methods.