The current implementation of Jaccard-similarity doesn't discard dupliate input values when calculating the denominator. I identified this issue by calculating Jaccard on identical input containing duplicates which didn't return 1.0 for which I added test cases.
When creating the test cases I forgot to set the call to Similarities::overlapSimilarity. Once fixed I saw that the calculation is also affected so I added a propsed fix for overlapSimilarity as well.
The current implementation of Jaccard-similarity doesn't discard dupliate input values when calculating the denominator. I identified this issue by calculating Jaccard on identical input containing duplicates which didn't return
1.0
for which I added test cases.