Jaccard Similarity

Jaccard similarity measures how much two sets overlap by dividing the size of their intersection by the size of their union. A score of one means the sets are identical, while zero means they share no elements. In AI and data analysis, it can compare document terms, customer interests, recommendation candidates, cluster membership, labels, or segmented image regions. The related Jaccard distance expresses dissimilarity as one minus the similarity score. The metric ignores repeated occurrences unless the data is represented differently, and it may be less informative when sets vary greatly in size. Choosing it requires confirming that shared membership, rather than magnitude or order, reflects the relationship that matters.