for a node X in a BBN is any evidence that enables us to update the prior probability values for the states of X. For example, if we know that the team is winning the match 3-0 at half-time, then the probability of win would be quite high, while the probabilities of both lose and draw would be quite low (compared with ignorance prior values).