Simhash and Minhash explained

The jaccard similarity between and ​​is given by

Cosine similarity is a vector space metric where similarity between two vector and is defined as where, is a dot product and is norm operator. Now, the tweets can be represented in user’s vector space as

and the cosine similarity is given as

Pairwise similarity scales quadratically both in terms of time and space complexity. Now, a k-dimensional minhash signature is defned by hash functions

Given , jaccard similarity of item & is defined as

where is an indicator function.

Source: mesuvash.github.io