It probably is using the term in context of Bayesian statistics, where it is a measure of how likely you estimate changes of something happening (the probability distribution), before you have any evidence on the phenomenon.