How to judge whether the similarity score is high or low?

Please see the example:

First

Then