Text Similarity

Hi Elif,

I have tried to use this activity but have gotten some conflicting results. Could you explain how you calculate the percentage? Here are some examples:

string1: "University of Boston"

string2: "Boston University"

similarity percentage: 15

string1: "are ana"

string2: "mere ana are"

similarity percentage: 50

I think the second example is ok, but I cannot deduce why the first example has such a low score. This activity would be perfect for getting a confidence score of synonims in a free text interpretation use case (think emails without a predefined data entry lists). Please advise.

Thank you,

Constantin