Concept - Jaccard Coefficient
Definition (s.35): popular similarity measure for vectors of asymmetric binary attributes, used with binary term occurrence vectors (1 = word occurs, 0 = absent).
= number of 11 matches / number of not-both-zero attribute values. 0-0 matches (both documents lack the word) are ignored.
Example (s.36)
d1 = “Saturn is the gas planet with rings.”, d2 = “Jupiter is the largest gas planet.”, d3 = “Saturn is the Roman god of sowing.” sim(d1,d2) = 0.44 (4 shared of 9 distinct), sim(d1,d3) = 0.27, sim(d2,d3) = 0.18.
Intuition
Ratio of intersection to union of the word sets. All words count equally (no TF-IDF weighting), so shared stopwords like is, the raise similarity; compare Concept - Cosine Similarity.
Source: 01 Intro - Chunk 02 (s.31-48)