You can think of the result of a bloom filter as a lossfully compressed index of the question: "Is this in the set?" which has the lossful property of sometimes

You can think of the result of a bloom filter as a lossfully compressed index of the question: "Is this in the set?" which has the lossful property of sometimes falsely answering yes. If you actually have the backing data then it's possible to re-compute a new result which excludes some items. It is also possible to confirm if an answer that was yes is actually in the set.