A Bloom filter represents a set using a bit array and k hash functions — inserting sets k bits; querying checks all k; false positives emerge as different keys hash to overlapping bits, revealing the probabilistic tradeoff.