well, actually spark is caused by many other things besides "short circuit" or "high potential".
Spark is caused by striking steel with flint.
Spark is caused by burning limb falling to ground.
etc....
as is SO common on these forums, the questioner did a very poor job of defining his problem and parameters.
if he truly only wants to detect -electrical- sparks, then UV-sensing is probably the most reliable. But if he wants to detect -any- kind of spark, including pure thermal sources, and do it without false-alarms from sun, electric lighting, people flicking a BIC, etc.....then it is a very tough job to do robustly.
I would look for info on flame-detection, as used for boiler start-up etc..