Okay, let's say I did this.
My plan was to install it in my beer fridge so I could keep an eye on my beer temp. If Coors Light drinkers can have their "certified cold" mountains, I should at least have a thermometer. Unfortunately this cheap little thermometer measured about 4°C too warm. And of course, since it costs less than a large coffee, there is no setting to calibrate it.
The probe is likely a simple thermocouple but could be a resistance thermometer - either way it's effectively a variable resistor affected by temperature. I figured I might be able to increase the resistance to get the results I needed.
I cut the lead and spliced in a 5KΩ trimpot (trimmer potentiometer) to one of the wires and dropped it in the fridge along with another thermometer that I trust. Waiting 10-15 seconds after each adjustment, I tweaked the trimpot until I was consistently getting the correct reading. With the trimpot removed, I measured it to be set to 2.2KΩ. I didn't have a 2.2KΩ resistor ( ▌▌▌) so I used a 1.2KΩ ( ▌▌▌) and 1KΩ ( ▌▌▌) in series. A little soldering and heat shrink and it's all good.
I should point out, this isn't really "calibrating". I just increased the resistance across the entire range - I'm betting it's not a linear relationship between the temperature and the resistance. However, it's now pretty accurate in the 0°C - 10°C range (above 10°C it starts to deviate). Good enough for a beer fridge.