I haven’t figured out or found on the net exactly what algorithms StumbleUpon uses, but I have found over the years that the correct measure for the similarity of two lists of things, e.g. websites two people liked, is the number of matches divided by the number of possible matches, which is the length of the shorter list:
Similarity = MatchCount / Length(ShortestList)
Also, looking for similar people, people with the same interests, or same kinds of page approval as you is all very well, useful for some purposes, but don’t expect to actually like or communicate well with such people. People with too much in common often disagree and often have communications problems.
What you really want to know are people similar to those who like and communicate well with people like you.