SimHash

SimHash is a technique in computer science for quickly estimating how similar two sets of data are, by producing hashes such that similar inputs get similar hash values, measured as the bitwise Hamming distance between them, rather than the all or nothing comparison an ordinary hash function gives. It was created by Moses Charikar. Google's web crawler uses SimHash to find near duplicate pages, and in 2021 Google said it intended to use the same technique inside its FLoC, or Federated Learning of Cohorts, system.