Hey Emilio, I think the example is right, let me explain. Token based similarity algorithms basically work with two sets of input, and the intention is to find similarity by comparing the elements of the sets. Now you can pass a single string or a set of strings, the code will work but the output’s interpretation will be different. If I am not wrong, you can confirm this by playing with the function, in case of only 1 string, each character is considered one element and hence the matching happens at character level.