3. (15 points) Consider computing an LSH using k = 160 hash functions. We want to find all object pairs which have Jacca
Posted: Sat Feb 26, 2022 11:18 am
3. (15 points) Consider computing an LSH using k = 160 hash functions. We want to find all object pairs which have Jaccard similarity at least t = 0.85. Suppose we use the (r,b)-way AND-OR construction, which means that a pair of documents with similarity s is considered as a candidate pair with probability 1 – (1 – 8). Choose the best r and b. Justify why your choice is the best.