Naive Bayes is a way of calculating the probability of something happening. It is especially useful for tasks like sorting emails as spam or not-spam or classifying text. It is called 'naive' because it makes a big assumption - it assumes that all of the pieces of evidence used to decide if something is spam or not-spam (for example) are completely independent of each other, which isn't always true. But even though it makes this assumption, it is actually really good at working out the probability of something happening.