Finds degree of similarity between strings, based on Dice's Coefficient, which is mostly better than Levenshtein distance.